HK1041121B - Touch-typable devices based on ambiguous codes and methods to design such devices - Google Patents

Abstract

The design of typable devices (6005), in particular, touch-typable devices (6005) embodying ambiguous codes (6001), presents numerous ergonomic problems. Solutions for these problems are herein disclosed. This invention teaches methods for the selection of ambiguous codes (6001) from the classes of strongly-touch-typable ambigous codes (6001) and substantially optimal ambiguous codes (6001) for touch-typable devices such as computers, telephones, pagers, personal digital assistants, smart cards, television set-top devices and other information appliances, given design constraints such as the size, shape, and computational capacity of the device, the typical uses of the device, and conventional contraints such as respect of alphabetic ordering or Qwerty ordering.

Description

typable device based on ambiguous codes and method

no marking

The invention relates to touch-key devices and to the use of such devices in calculators and electronic communications. And particularly to touch-key devices designed based on easy-touch ambiguous codes and optimal ambiguous codes.

Since the invention of typewriters 100 years ago, keyboard engineering has become a fairly active area of development, resulting in many competitive designs. With the growth of personal computers and electronic communications, designers have come to adapt to various kinds of constraints and to grasp opportunities that are brought about by this new technology, so that the number of keyboard designs has increased substantially. However, the diversification of these prior art keyboard designs is not due to the different kinds of limitations and opportunities that arise, rather, as a result of keyboard designers not having a complete understanding of existing limitations within the problem they are addressing. Also, the lack of an overall, efficient method for optimizing these limitations is reflected. Therefore, the situation of the technology at the present stage has too many incomplete solutions. These disadvantages are completely overcome by the keyboard design method of the present invention. To demonstrate the invention, the optimization method will be applied to the design of multiple device materializations, each of which is the best basic solution under a series of design constraints.

The present invention relates to touch-key devices. Touch-typing is a rather difficult hand skill, just like playing a musical instrument. Once learned, it is difficult to correct the formed movement pattern again. This places considerable constraints on keyboard design. The well-known Qwerty keyboard (and its close variants such as the Azerty keyboard commonly used in france) is greatly advantageous because of the deep stalk and excessive learning of the typing movement patterns. Thus, the broad base of the Qwerty keyboard makes learning more advanced keyboards, such as the Dvorak keyboard, difficult. Although Dvorak keyboards also have their users, the number is not large. The Qwerty keyboard is not suitable for a palm-type or small-type typing device because of the large number of key buttons. The advent of palmtop or small typewriter devices brought a line of chance to keyboard designers. Since the existing repetitive movement pattern tends to be fixed, the new design appearing in the moment can keep its dominance in the future if it is dominant, and even in this field, new designs are continuously appearing. This places a significant burden on the keyboard designer to avoid having to use a suboptimal design for future keyboard users.

In this prior art, instead of editing a series of symbols, there are two main methods that can reduce the number of typing methods 1) the coordination method, the combination of typing methods editing each symbol, 2) the ambiguous code, each typing method editing the combination of symbols. Coordination is not practical because learning to operate coordination is difficult and there is little time one would like to invest in. Therefore, only ambiguous codes, or a combination of ambiguous codes and a coordinated approach, can give a true solution to the problem.

The present invention is directed to a method for designing an optimal ambiguous code in a typing device.

It is a further object of the present invention to provide a method for designing easy-to-touch ambiguous codes in a typing device.

It is a further object of the present invention to provide a keyboard suitable for touch typing on both conventional and miniature keyboards.

It is a further object of the present invention to allow information from alphanumeric cells to be transmitted from a conventional telephone or two-way pager to another device of the same type without human intervention, and thus to be less expensive to service.

It is a further object of this invention to provide a touch-typable personal digital assistant.

It is a further object of the present invention to provide a keypad for a driver of a vehicle that can be keyed in without distracting the driver.

It is a further object of the present invention to provide a low cost key-in communication device for a manufacturer that can be integrated with a conventional telephone communication system.

Some preferred embodiments of the present invention are directed to assisting a typist trained with a conventional keyboard to shift his or her typing skills to accommodate a new keyboard by retaining some of the design of the conventional keyboard on the new keyboard.

It is a further object of this invention to provide an integrated method for producing ambiguous codes with minimal search error rates.

It is a further object of this invention to provide an integrated method for producing ambiguous codes with minimal query rates.

It is a further object of this invention to provide such a device that reduces typing injuries.

It is a further object of this invention to provide a twice-foldable palm-size calculator device.

It is a further object of the present invention to provide a keypad for one-handed use of a device in a palm-top calculator.

It is a further object of the present invention to provide a one-handed and two-handed keypad suitable for use with a palm-top calculator or desktop keypad.

It is a further object of this invention to provide an easily learned coordinated keyboard.

It is a further object of this invention to provide a synergistic hybrid keyboard.

It is a further object of the present invention to provide a touch-typable query mechanism for ambiguous keyable devices.

It is a further object of the present invention to provide a touch-based discrimination elimination model for typable devices having ambiguous codes.

It is a further object of this invention to provide a hybrid harmonized/ambiguous code keyboard that can be fully integrated with standard telephone systems.

It is a further object of this invention to provide ergonomic designation of symbols for the patterns. It is a further object of this invention to provide a transparent touch interface for a typable device having a touch screen.

It is a further object of this invention to provide optimizations across a set of natural languages.

It is a further object of this invention to provide a key-able device that can be held with a single hand and that has a reduced scanning time.

Other aspects of the invention are described in detail in the following paragraphs.

Brief description of the drawings

A preferred embodiment of this invention will now be discussed in more detail with respect to the drawings, the following of which is a brief description.

An overview of the optimization considerations for a keyable device made in accordance with the present invention is shown.

FIG. two shows a flow chart of the construction of a device fabricated based on the easy-touch ambiguous code.

FIG. three shows a flow chart of construction of ambiguous codes that meet at least one ergonomic criterion and are optimized with respect to those ergonomic criteria.

Figure four shows a flow chart of a specific embodiment of the method of figure three using a method of on-demand optimization.

Figure five shows the distribution of the probability of finding errors for some of the ambiguous codes on the selected key.

Figure six shows the distribution of query probability for some of the discriminatory on demand codes on selected buttons.

FIG. seven is a flow chart showing guided random walk optimization.

FIG. eight shows a flow chart of the construction of easy-touch ambiguous codes.

The ninth graph plots the search error rate versus the number of on-demand buttons, and the optimized ambiguous code.

The plot ten plots the query rate versus the number of on-demand buttons, and the optimized ambiguous codes.

FIG. eleven shows the optimization of the search error rate versus query rate for ambiguous codes over a range of key button numbers.

FIG. twelve shows a chart relating to the hierarchy of key numbers accessible to the hierarchy under several different optimization methods to achieve the hierarchy.

FIG. thirteen shows a flow chart of a method of synthesizing encoded symbols.

To assist the reader in understanding the uniformity of the device embodiments of this invention, the table of the fourteenth figure is an abstract of these embodiments and features thereof. These embodiments are provided to clearly and unequivocally teach the broad scope and various aspects of the invention.

Fig. fifteen shows an embodiment of a smart card with 16 keys for coding of alphabetic characters.

Fig. sixteen shows an embodiment of a smart card with 9 keys for the encoding of alphabetic characters.

Fig. seventeen shows a keyboard mounted on the steering wheel.

Fig. eighteen shows 10 keys in the phone with optimized codes.

Nineteenth of the figures shows that the ordered ambiguous codes in alphabetical order of reduced ambiguity apply to cellular telephones.

The diagram twenty shows a Qwerty-like keyboard with optimized lookup error rate and query rate and related to the arrangement of letters in each row of the Qwerty keyboard.

FIG. twenty-one shows another Qwerty-like arrangement of the keyboard.

Figure twenty-two shows a divergent keyboard with a standard numeric keypad arrangement.

Twenty-three shows a discrimination elimination mechanism with ergonomic touch-key guidance.

FIG. twenty-four shows a method flow diagram. The method answers the query in a touch-key guided manner.

Figure twenty-five shows a one-handed keyboard design. This design preserves the typing skills between single-handed and two-handed keyboards.

Figure twenty-six shows a design of a two-handed keyboard. This design preserves the typing skills between single-handed and two-handed keyboards. In this case, a two-handed keyboard is selected for the greatest similarity in typing status in both types of keyboards.

Fig. twenty-seven shows a design of a two-handed keyboard. This design preserves the typing skills between single-handed and two-handed keyboards. In this case, the two-handed keyboard can be equally distributed between the two hands.

FIG. twenty-eight shows a keyboard in combination with a mouse.

Twenty-nine shows the front side of a twice-foldable information device in an unfolded state.

FIG. thirty shows the back side of a twice-foldable information device in the unfolded state.

Thirty one shows a twice-foldable information device in a state of being folded once, showing additional functions thereof.

Fig. thirty-two shows a twice-foldable information device in a state of being folded twice, showing another function.

Thirty-three shows a two-fold information device in a separated state, which can be typed with both hands.

Thirty-four shows a typical personal digital assistant with a touch screen.

Figure thirty-five shows a typical personal digital assistant with a potentially transparent keyboard.

FIGS. thirty-six A, B and C show three modes of a 16-key keyboard.

Figure thirty-seven shows a standard telephone arrangement.

Figure thirty-eight shows a hybrid harmonization/ambiguous code keyboard programmed into the phone.

The thirty-nine plot shows the distribution of the search error rate and query rate for a mixture of all harmonized/ambiguous codes for a particular structure as compared to the search and query rate for a standard ambiguous code.

FIG. forty shows a flow chart of the construction of multilevel easy-touch ambiguous codes.

FIG. forty-one is a flow chart showing the specific implementation of multilevel easy-touch ambiguous codes.

Fig. forty-two shows a typable device suitable for use with the multi-level ambiguous code of fig. forty-one.

FIG. forty-three illustrates the operation of the forty-two devices showing the first level of multi-level ambiguous codes.

FIG. forty-four shows a second level of coding for a multi-level ambiguous code.

FIG. forty-five illustrates the operation of the forty-two device showing portions of a second level of encoding of a multi-level ambiguous code.

FIG. forty-six shows the sequence of operational states for the forty-two devices when they use multiple levels of ambiguous codes to type "think".

Forty-seven, like forty-six, shows the sequence of operational states for the forty-two device using a multi-level ambiguous code to type "think". However, in this case, the operation of the visual cache to reduce the scanning time is also shown.

Detailed description of the invention

Definitions and basic concepts

This section is intended to define the words and concepts that will be used in the following detailed description.

A language is a set of symbols that an individual can establish and determine the likelihood of order. The symbol groups, symbol sequences and the possibility of these sequences are here languages. For clarity of discussion and without limiting the scope of the invention, the language in which we refer to is written natural language, such as English, although in particular, we may refer to the symbols as "letters" or "punctuation", but with the general skill of the art, the symbols discussed herein may be any unassociated written unit, including standard symbols, such as ideographic characters of Chinese, or symbols created by the rename of the singer "prince".

Keyboard/typing methods a keyboard is a component of a communication and/or computing device. The keyboard translates the limb movements of the operator into a sequence of symbols. The keyboard is composed of at least one typing mode which is responsible for the translation of the subset of physical actions operated to activate the subset of symbol sequences of the keyboard.

The limb action to operate the keyboard is typically the movement of a finger and/or thumb or a palm-type stylus. This definition also extends to other body actions such as movement of the head, tongue, or eyes. These movements may be used to signal symbol selection from the keyboard. The device with the keyboard definition is the typable device.

It is understood that "key-able device" is not only a device having a keyboard, but that the entire communication system is embedded within the key-able keyboard. The limits of this system are defined by the basic ambiguous code architecture. If the entered symbol in a typable device appears directly on the display screen, which is a part of the typable device, the limits of the system are quite clearly defined by the size of the device. In more general terms, a key-able device includes a microphone which transmits information to the central computer which is responsible for encoding or processing the original information from the microphone. Such typable device of course comprises such a central computer and its operating mode is set by the software established according to the content of this invention.

At least the way of typing in each keyboard requires a great variety of physical manifestations. The key type method is mainly characterized in that it allows an operator to select a subset from a group of symbols to make the keyboard encoded. With this layer of knowledge, and to increase the readability of this specification, the "key buttons" are often interchanged with the "key types".

Typing is the process of sequentially selecting at least one type entry in order to select a sequence of symbol subsets from a set of symbols encoded by a keyboard. Known handwriting recognition software allows typing to translate a sequence of pictorial actions into a sequence of subsets of a set of symbols.

The flow of touch typing symbol sequences generated by the keyboard mostly or only uses kinesthetic feedback rather than visual or auditory feedback.

Extremely relevant symbols and symbol sequences are known, as are the frequencies of occurrence of different letters in a character. For example, in the previous sentence, the letter "e" appears 11 times, and the letter "z" does not appear once. This situation also occurs with two alphabets, three alphabets, and so on. Quite clearly, the frequency of occurrence is different for all characters. The three-character word "the" is common in English, while the three-character word "zap" is quite uncommon. These statistically irregular variations can be used in the design of ambiguous codes. Indeed, at least since the invention of the Qwerty keyboard, statistical irregularities have often been used in keyboard design.

We are also concerned particularly with certain symbols and symbol sequences whose distribution in a typical sample of text content has a great correlation with other symbols or symbol sequences. Such symbols are referred to as extremely correlated symbols. For example, the symbol "is used in English and other languages to represent the end of a sentence, which may be a highly relevant symbol because the distribution of sentence lengths within typical text content is not random. In hebrew, the symbol "is also correlated to a particular alphabetic symbol, although hebrew uses a different symbol for the letters at the end of the word, and the end of the sentence is also correlated to the end of the character.

The reference data is used to measure the symbol sequence of the symbol correlation, and the reference data is generally estimated by a reference corpus analysis. A corpus is a collection of a relatively large amount of text, with the selected text representing some portion of the language. It is well understood by linguists that there are many fundamental problems in creating a corpus that represents the entire characteristics of a language, sometimes blocking characteristics associated with a particular type of textual content or a particular type of author. These problems have exceeded the scope of this invention. Here we take reference data gathered from the english national corpus from the beginning to the end. The english national corpus is the largest existing corpus for analyzing english. The selection of a corpus is a necessary step in the gathering of results, allowing comparison of various methods and embodiments. The choice of choice is not to be construed as limiting the scope of the invention. Particularly, the selection of English text content and a material library is a judicious choice. The same analysis can be used for any other written natural language.

In the united states, the keys of a telephone keypad are usually labeled with both letters and number elements, and the key of number 2 is labeled with a, b and c; the key button of the number 3 is also labeled d, e and f, and proceeds in alphabetical order.

Thus, the key button sequence of 233 is also followed by the alphabetic sequences add, be and bed, all in English, although other nonsense alphabetic sequences such as cff may also be produced. If the order appears in a meaningful sequential reference list, the order is considered to be meaningful. These alphabetic sequences, whether meaningful or not, are therefore associated with the same numerical sequence. We refer to the key sequence of the key 233 as the code, and the sequence of add, be, bed, eff, etc. is the decoding of the code 233. To solve the confusion, "decoding" refers to "meaningful decoding". Taking this as an example, the set of symbols used for decoding are either decoded symbols, or just symbols, in alphabetical order, if not for any confusion. And the set of symbols used for decoding, for example, a number, is considered a decoded symbol.

Ambiguous codes are well known in the art. On the standard telephone keypad in the united states, there are a total of 12 keys, 10 of which encode a number and several of which encode 3 or 4 letters, arranged in alphabetical order in english. These assignments yield an ambiguous code, which we call the standard ambiguous code. This is encoded as abc def ghi jkl mno pqrstuyv wxyz.

Since each button is encoded with several letters, a disambiguation method must be used to determine which letter type operator wants to use. In general applications, such as voice messaging systems, the decision to use a letter is made by comparing the order of typing with a stored list of replies. If several replies in the stored reply list are in accordance with the typing sequence, the operator must select them by himself. The display of the order within the selections may be at the discretion of the individual or may be at the discretion of which reply is the correct frequency of replies, and the replies are presented in the order of decoding of the frequency.

The standard keyboards are widely used and there are three main types of standard keyboards: the Qwerty keyboard and its similar variants, the 12-key phone keypad, and the 17-key number keypad and its similar variants. This invention has the unique advantage of providing a useful method for a standard telephone, a number keypad, and a specially designed keypad not covered by this description.

The symbol designation of the key-easy degree non-key-easy type device, namely the key button in the device is already fixed; only typists familiar with the device can develop the instinctive response of the limbs, using a particular movement pattern to decode a particular symbol. The device with easy touch key degree has the following three characteristics: 1) low touch typability, 2) based on ambiguous codes, 3) in a normal mode of operation, a touch typist can type text content with an acceptable degree of accuracy using a typable device without undue distraction for the touch typing task to interfere with the ambiguity resolution process.

The degree of easy-touch keying is a matter of degree; it is a measure of the degree of touch. The degree of touch depends on many factors, some of which are personal problems for the typewriter, some of which are used with the typable device, and some of which are built into the typable device itself. For example, a typable device may be sufficiently touch-sensitive for a touch typist to handle some typing tasks, but not others.

Two key definitions of easy-to-touch, text content accuracy and distraction of touch typers depend on many factors, including:

□ A method of disambiguating a medical device,

□, when the machine is used, for example when driving or sitting at a table,

□, the type of text to be typed, the text determining the degree of precision necessary,

□ the reference data is recorded on the recording medium,

□ the skill of the typist,

□ personal preferences, an

□ the disambiguation mechanism attracts the user (e.g., when the speech synthesis mechanism speaks a sentence or question, the mechanism may or may not be distracting to the user from a ring or flashing light).

Although the degree of easy-touch bonding is a matter of degree just like temperature, it is also just like temperature, and its definition is very clear. Once these various factors have been determined using standard experimental protocols well known to those skilled in the art, the degree of accessibility of a key-able device can be quantified in relation to the user or user of the finger cluster. More specifically, two portions of the degree of easy-touch can be measured in ambiguous codes: searching for errors and querying. Therefore, the value of the degree of easy-touch can be specified without directly referring to any number of users, only the ambiguous code described above.

Like temperature, the degree of easy-touch bonding has a lower boundary. It is clear that for any typist, if a device must intervene to eliminate ambiguities after every character entered or every three characters entered, the device cannot be said to be a touchable device. The lower boundary of the degree of easy-touch keying can be presented by the persistence of attention. If a typist's attention must always be focused on the operation of the ambiguity resolution mechanism to produce an acceptable level of text, the device is not touch sensitive.

The actual lower boundary for typographers with average degree of typographical skill is associated with the theoretically lower boundary mentioned above. In order to make the inventive concept of the degree of easy touch accurate and precise in both figures and concepts, the numerical value of the degree of easy touch is presented by looking up the wrong and queried values. This numerical characterization more clearly indicates that the methods and apparatus of the present invention differ from the past.

The easy-touch degree ambiguous code is the basis of the easy-touch degree device.

Feedback device this device allows the user to intervene at different points to decode the sequence of symbols generated by the ambiguous code, the internal feedback of which to the user's sensory perception is necessary. Usually this feedback is presented in the form of a graphical symbol, but feedback can take many forms, such as auditory, tactile, and even olfactory, etc.

Ergonomic considerations relate to the design of keyboards with ambiguous codes and the elimination of limitations. This may include a reduction in the search error rate, a reduction in the query rate, a number of keys that the keyboard size fits into, a degree of integration with existing keyboards such as Qwerty keyboards, telephone keypads, or numeric keypads, stability of the spacer structure, structural accuracy, minimal usage of mode shift keys, spacer structure, integration between one-handed and two-handed typing, and conventions that preserve, for example, alphabetic ordering. Other limitations include: the ergonomics of the disambiguation mechanism, the ergonomics, look and feel of decoding of symbols of interest, and the computer resource availability at the transmitting and receiving end of communications utilizing the ambiguous codes.

The lookup error measures the error generated by the ambiguity resolution mechanism that systematically selects the most probable (most significant) decoding from a series of probable decodes in an ambiguity sequence to resolve an ambiguity. Thus, the search error rate for an encoding is the total number divided by all possible decodes that are most likely to be decoded in an order other than the ambiguous sequence, which is the reference likelihood for the possible decode. In the case of character-based disambiguation, these sequences begin and end with a "space key". That is, these orders are characters. A lookup error refers to the likelihood that the most likely decoding is not correct. The lookup errors are presented in a rate manner. The lookup error rate is the ratio of characters to lookup errors. The lookup error rate is the inverse of the likelihood of a lookup error.

The query likelihood is the total number divided by all non-unique (meaningful) encodings, which are the reference likelihoods for the aforementioned encodings. This indicates the likelihood that more than one meaningful interpretation of a character may be possible. Therefore, the user must use the query to determine which decoding to use. The inverse of the likelihood of a query is the query rate, in terms of the ratio of characters to the query. The query rate provides the average number of characters typed between queries.

Substantial optimization we call substantial optimization if a code is one of the best codes in terms of performance, taking into account other constraints imposed on the coding. For example, a code on 20 keys may have a lower seek error rate than a code on 2 keys, but the code on 2 keys may be optimized for the seek error rate given the limitations of the code on 2 keys. An optimized ergonomic code may be defined as a code that is optimized for each set of ergonomic constraints at the same time.

These limitations include, but are not limited to, the number of buttons, the lookup error rate, and the lookup rate. These three constraints, one set of constraints for each group, are related. The lookup error rate tends to increase with the query rate, while the lookup error rate and query rate increase as the number of buttons decreases. When a given criterion is the only optimization criterion, and some other criteria must be optimized, the best possible value of the given criterion may be better than the best possible value that can be obtained. Thus, the ergonomic limitations associated with known designs and their importance must be considered as the initial steps of the optimization method proposed by this invention.

It must be emphasized that there is no way to positively discuss the optimization of ambiguous codes, but correlation with the reference data set of the language to be coded must be evaluated. Indeed, in view of any ambiguous code, it is possible to create a data set and optimize the coding for the data that has been created.

An estimate of the optimization of a given code may be obtained experimentally from the random code, taking into account the reference data set. The random encoding will be discussed in detail later.

The ambiguity resolution method must be in terms of the selected ambiguity resolution method if the substantial optimization of the ambiguity encoding is to be properly defined. An optimized code associated with one ambiguous solution may not be optimized for another ambiguous solution code.

In this art, there are at least two well-known methods of disambiguation. These are character-based and block-based disambiguation methods. In character-based disambiguation, a table of characters and their possibilities is used to select between alternative translations of known codes in the ambiguous code. For example, all characters known to be meaningfully decoded in the encoding will be compared, and the word with the highest probability will be selected. Block-based disambiguation methods are also quite similar, except for the possibility of containing segments of text content in the list, and the likelihood of segments that are of a certain size.

Word-based and block-based disambiguation methods are special cases of the general architecture. The general architecture is a sequence-based disambiguation method in which a sequential list is associated with probabilities and disambiguation is performed from the list. It should be noted that in languages such as English, the "blank" symbol defining the boundary of a word is for the purpose of this discussion and is not different from other symbols. An individual may define a list of sequences and sequence capabilities, where the sequences include "blank" symbols, and thus extend beyond the boundaries of the words. The individual may further define an order comprising wildcards and may thereafter define an order list comprising any sub-sequence that does not necessarily correspond to a word within the language. In such a case, a linguistic arbitrary compound representation can be created and used in the discrimination elimination method.

For example, grammatical and semantic relationships between subsequences can be used to eliminate conflicts between possible interpretations of ambiguous code sequences. In order to make it more readily apparent, we will focus on the well-known character-based disambiguation method for this application, except where specifically indicated. Those skilled in the art will appreciate that the method taught by the present invention is not necessarily character-based disambiguation methods, and any other disambiguation method may be used.

A partition separated by an integer n is a set of integers when their sums add up to n. Typically, a known integer will have several divisions, for example, 5 may be divided into 3:2 and 2:2: 1. Those skilled in the art will be familiar with the methods of reciprocity used to generate all the possible separations of an integer. Most conventional coding uses a partitioning method that is as even as possible. That is, in a partition, the number of letters on each key is the same, within a possible range, depending on the number of letters to be encoded. This choice will be explained more in detail below as a reasonable choice in some ergonomic considerations, although this may be a less important consideration in other respects.

There are two pluralities of ambiguous codes that are claimed herein for their exclusive use. The two pluralities are 1) the easy-touch ambiguous code, and 2) the substantially optimal ambiguous code. The ambiguous code may be substantially optimal but not touch-sensitive, or touch-sensitive but not substantially optimal, or neither substantially optimal nor touch-sensitive, or both.

The discovery begins by indicating how to find the ambiguous code in the two complex numbers and in which complex number to find the code. The following is an explanation of how these codes in these two pluralities can be used to make typing devices, and how these codes can be used to solve different design problems encountered by designers of typing devices.

The best mode of the invention is set forth in the combination of design criteria which should be optimized in accordance with the teachings of the present invention. The series of methods and devices proposed by the present invention will now be described in several practical and practical contexts.

The variety of devices that can be constructed by one of ordinary skill in the art in light of the teachings of this invention will exceed the number of applications selected herein. Many different extreme or special cases that are subject to design constraints may be addressed in this application. It will be apparent to one skilled in the art, given the guidelines contained in these cases, how to combine these features appropriately to solve the overneutralization or synthetic design problem.

One such application is optimization with reference to the lookup error rate alone. This application is designed for a device with small memory and low computer functionality, an example being a smart card. Such an arrangement, the computer resources may not be available to support a compound query process for the user during the disambiguation process. The apparatus thus uses a simplest ambiguity elimination procedure, namely a systematic selection of the most likely decoding of any given code.

Another application is to optimize with reference to query rate alone. This application is designed for the driver of a vehicle, an automobile being an example. Even if the computer resources are sufficient to support the complex query process, the use of such a process should be minimized in order to minimize driver distraction from driving.

The next application is to provide phone keypad reference lookup error rate and query rate optimization while being compatible with common phone keypad designs.

Yet another application is to optimize with reference to customary criteria: arranged in order of saving English letters. The letters are arranged in order on the keypad of a typical touch-type telephone. By optimizing the separation, it is possible to preserve the alphabetic ordering in a conventional telephone keypad while reducing the search error rate and query rate.

Optimizing the partitions provides more of the substantially optimal query and lookup error rate keypad shown and preserves the application of the traditional Qwerty keypad arrangement as much as possible.

A further application illustrates that a keyboard design that is as consistent as possible with conventional designs is a keyboard that is built on ambiguous codes and that conforms to a number keypad.

In many applications, an ergonomic keyboard that can simultaneously perform divergent and divergent cancellation operations would be advantageous. So far, it would be preferable to choose the ambiguous code among a number of keys that divide the desired code symbol. A sign number key number equal to 1/2 is most commonly used. The ideal result of this common choice will be shown in the next application.

Another application is how the keyboard can be optimized for cross-platform compatibility. In this application, two keyboards, a single-handed keyboard and a two-handed keyboard, are designed to operate at high speed transitions, so that the touch typing action for operating one of the keyboards is transferred without difficulty to the touch typing action for operating the other keyboard. This keyboard has the additional advantage of reducing the likelihood of typing injuries, among other objects and advantages, as will be explained in the detailed description.

The integration of all the above applications illustrates that the use of different keyboards represents different degrees of optimization, while a given user may need to use the keyboard in several different situations, which also represents the necessity to have the program co-exist with different solutions on a single device. There is a surprising solution to this problem because of the size of the miniaturized typing device that can only be achieved with ambiguous codes, i.e., the bifold pda referred to in this application.

The several applications discussed so far are related to hardware and software specifications. However, it is possible to use a completely or mainly software-based solution to achieve many of the objects of the invention. An example of software processing will be described in detail herein to explain how the use of appropriate software can combine the features of existing hardware to achieve some of the goals of the invention.

The last group of applications has happened unprecedentedly to combine two alternative ways to make low key button keyboards: coordination and ambiguous coding.

First, this illustrates that the ergonomically built coordination model, in combination with the optimization of query rate and search error rate, can make the ambiguous encoding function on an n-key similar to the ambiguous encoding function on an m-key that is substantially larger than the n-key. When the method is applied to the generic ambiguous codes, the 8-letter keys on the generic ambiguous codes result in substantially optimal code properties without requiring coordination over the 13 keys. In all of the remarks and discussions herein regarding how to extend ambiguous codes, reference is made to English, but the remarks and discussions herein apply to other languages as well. In particular, while discussed herein with respect to this application, the headnotes are generally applicable to all applications.

Secondly, this illustrates that by combining a split-seize approach with the exemplary approach in the aforementioned application, the number of typing devices can be further reduced, in this example using 4 typing devices to operate a 16-code ambiguous code device. The number 4 is selected so that a hand-held device applying the code can be operated by the index finger and thumb of the hand holding the device.

Summary of the operation of the easy-touch key device.

FIG. 2 illustrates an operational overview of a touch-enabled device based on ambiguous codes. Such devices have a keying function and operation of the device (step 140) generates a sequence of code symbols (step 141). The touch sensitive ambiguous code is used to locate the encoded symbol sequence and the decoded symbol sequence in step 142. These encoded symbol sequences are then optionally output to a display on which the user of the device can directly view them, or electronically for further processing, transmission or storage (step 142).

It is specifically noted that the ambiguous code device of the schematic representation of fig. 2 meets other ergonomic criteria besides the degree of easy-touch keying.

Establishment of substantially optimal code the steps of a method of studying ambiguous codes that optimize a reference to a set of ergonomic standards are explained with reference to FIG. 3. Overall, these steps are as follows:

□ step 2000 includes the following sub-steps of selecting a group of statistically related decoded symbols to be embedded in an ambiguous code

2007 selects a set of reference data to be used,

2008 selects a disambiguation method in accordance with the statistical correlation of the data analysis symbols selected in step 2007 in step 2001.

□ step 2002 selects an encoding symbol number.

□ step 2003 selects ergonomic criteria based on the code that should be substantially optimal.

□ step 2004 compares the importance of the ergonomic criteria selected in step 2003.

□ step 2005 selects an optimization method.

□ step 2006 applies the optimization method selected in step 2005 and thereby produces substantially optimal ambiguous codes.

It can thus be seen that steps 2000 through 2003 can be applied in any order, and the application of one of these steps will be able to influence the selection among the other steps. The details of the application of these steps will now be described.

Step 2000 selects a group of statistically correlated decoded symbols to implant an ambiguous code. This step also includes substep 2007 of selecting a set of reference data, and substep 2008 of analyzing the statistical correlation of the symbols according to the data selected in step 2007. The purpose of these steps is to find those symbols that can be implanted differently. All ambiguity resolution procedures use the correlation between symbols to predict which decoded symbol sequence should be concatenated with an encoded symbol sequence. If a decoded symbol is randomly distributed in the text to be encoded, it cannot be implanted in an ambiguous code because it is impossible to predict a randomly distributed symbol. Generally, for any natural language, the symbols used to encode that language (e.g., letters in English and ideograms in Chinese) will be sufficiently statistically correlated to facilitate the design of efficient ambiguous codes for those symbols. There are also other symbols, such as punctuation, that may be sufficiently statistically related, both between these symbols and between the letters or ideograms required for writing the language. The particulars of steps 2007 and 2008 will depend on the language in which they are intended to be represented. Methods of analyzing the statistical relevance of symbols used in natural language text are well known to linguists.

In step 2001, a discrimination elimination method is selected. As mentioned above, there are at least two well-known types of disambiguation procedures, block-based disambiguation procedures and character-based disambiguation procedures. Both methods use statistical correlation between symbols to predict which decoded symbol sequence should be concatenated with an encoded symbol sequence. Both block-based and character-based approaches can be augmented by using high-level information about the language, such as its grammar and semantics. The purpose of these steps is to create an ambiguous code based on the selected ambiguous cancellation procedure, thereby associating the most selected decoded sequence with each encoded sequence. The details of the selected disambiguation procedure may affect the detailed nature of the ambiguous code designed by the method. The method will be described with the option of character-based disambiguation as the disambiguation program, although other disambiguation programs will be discussed.

Step 2002, an encoding symbol number is selected. The selection of the number of code symbols is of great importance for the design of typing devices based on ambiguous codes. This choice is made by considering many factors, including the size of the typing device and the acceptable degree of discrimination. These factors and their interactions are best explained by reference to practical examples, which are used hereinafter.

In step 2003, ergonomic criteria are selected based on the code that should be substantially optimal. One of the key points of the present invention is to discover and define several ergonomic standard ambiguous codes that determine the quality of the typing device. These criteria include key-touch sensitivity, lookup error rate, query rate, structural accuracy, physical accuracy, preservation of traditions, partition structure, cross-platform compatibility, design rule sensitivity, and scan speed. Depending on the application, more than one criterion may be applicable to the design of the typing device.

In step 2004, the ergonomic criteria selected in step 2003 are compared for significance. While more than one criteria may be appropriate for the design of the typing device, some dependence on these criteria must be made depending on their importance. It is rare to achieve the same optimum for a given ergonomic criterion when this ergonomic criterion is considered alone and when it is simultaneously desired to optimize it against another ergonomic criterion.

In step 2005, an optimization method is selected. Two optimization methods, random selection and guided random advancement, are discussed in detail below. Random selection between these two is generally easier to apply, while guided random advancement is able to produce better codes. These two optimization methods are representative of many possible methods that may be suitable for a given typing device design. In some cases, such as the first harmonized/ambiguous encoding device considered below, the number of codes that need to be examined is small enough to be a thorough comparison of each.

At step 2006, the optimization method selected at step 2005 is applied, and thereby yielding substantially optimal ambiguous codes. Regardless of the optimization method selected in step 2005, certain techniques must be employed in applying the method that produces the substantially optimal ambiguous code. Especially when an optimum is needed immediately over several ergonomic criteria, it is preferred to consider each ergonomic criterion separately, so that a prediction can be made of the finally achievable coding quality. This prediction may be considered invaluable for fine-tuning optimization, as will be discussed in more detail below, once the two optimization methods are fully described.

The basic method of randomly selecting a code with good properties is to randomly select a code, test the properties of the code, and select the code with the best properties. Exhaustive enumeration, i.e., testing all codes in the candidate block, is generally not a viable option since the number of codes to be tested is too large for any reasonable computation time.

The random selection provides a rod according to which the functionality of other code selection methods can be measured. Assume that a set of ergonomic criteria and a comparison of such criteria are known. We can predict the substantial best for a first ambiguous code by randomly choosing to generate additional ambiguous codes, referring to this criterion and its comparison. If it is possible to find a code with a value equal to or better than the first code in a small random test with reference to known ergonomic criteria, the first code is not substantially optimal.

Conversely, if a substantially greater number of random tests are required to generate codes having values that are better or equal to the first code, or if a better code exists, then the first code is not substantially optimal.

Referring to FIG. 4, we detail a method for eliminating the hypothesis that candidate ambiguous codes are substantially optimal. Overall, these steps are as follows:

□ step 3000 determines a set of correlation constraints that define a suitable code set containing candidate codes.

□ step 3001 determines an ergonomic set of criteria under conditions where the candidate code may be substantially optimal.

□ step 3002 randomly selects a subset of codes from the set of codes determined in step 3001.

□ step 3003 evaluates the individual code selected in step 3002 against the ergonomic criteria determined in step 3001.

□ step 3004 compares the candidate code values according to the ergonomic criterion determined in step 3001 and the values obtained in step 3003. If any of the values obtained in step 3003 are better than the value of the candidate code, the assumption that the candidate code is considered substantially optimal may be eliminated. The details of these steps are as follows:

step 3000 determines a set of correlation constraints defining a suitable code set including candidate codes. The code block must be properly defined in case a substantial optimum of a candidate code needs to be evaluated. Some possibilities are associated with the limitations: the number of code symbols, the separation structure and the recognition of a particular arrangement, such as the sequential arrangement of letters. Each constraint limits the code set that has been properly compared to the candidate code.

At step 3001, a set of ergonomic criteria is determined if the candidate code is substantially optimal. Some of the criteria that may be relevant to the analysis of candidate codes are: error lookup error rate, query rate, recognition of a particular arrangement, such as alphabetic ordering, recognition of a custom design and structural accuracy.

Once steps 3000 and 3001 are performed, the distribution of the code-block coding properties is defined and the values of the distribution can be randomly sampled. Taking fig. 5 as an example, the block codes can be defined as 1) as equal a separation as possible and 2) the number of symbol codes designated by 7, 9, 11, and 13, respectively. The above sum is defined as step 3000. Completing step 3001 confirms that the lookup error rate is the only ergonomic criterion. Steps 3002 and 3003 are combined to obtain a distribution of values, and the pattern of the distribution can be obtained by random sampling. Taking fig. 5 as an example, 5000 codes are selected from each distribution as step 3002, and the search error of each code is measured as step 3003. This value is expressed as a ratio of the percent of search errors (the inverse of the search error rate) to the encoded number of known error percentages. The peak of the distribution pattern becomes more pronounced as the number of key buttons increases. If the above steps are repeated with the search likelihood instead of the search likelihood, the data in FIG. 6 can be obtained.

To illustrate step 3004, a candidate code to be tested must be selected. This code is a 14-key code such as the proposed pn gt cr zk wj a e hi so ud xf ym vl qb [1 ]. The comparison of the encoded lookup error with the reference data is such that there is one lookup error per 105 words. Proceeding according to the above steps, we find that 14-key codes separated by as equal a spacing as possible can be obtained in an average of seven random tests with the search error equal to or better than the candidate code. If we repeat the above steps with query likelihood as an ergonomic criterion instead of search likelihood, we find that having a better code than the candidate code query rate (query every 4 words) will occur on average three times every four times in a random test. Thus, the known ambiguous codes of [1] are substantially optimized in terms of either lookup error rate or query rate. Indeed, most of the encoding is better than this known encoding in terms of query rate.

Generally, if a code has not been optimized according to ergonomic criteria, it is likely that the code is not substantially optimized under proper linguistic data metrics.

Guided random walk

Guided random marching is a repetitive optimization method. The method has new codes generated at each step based on the best code from the previous step, and these new codes may be better than the best code from the previous step. As the steps are repeated, better coding results. This set of procedures will be explained slightly herein and discussed formally further.

In the present case, we lack this knowledge with respect to the optimization of ambiguous codes for more than one ergonomic standard, and the continuation of the best-finding direction is blinded to anyone as it appears to be. The safest approach is then to take the smaller and better steps from the more and better directions, and never to take other actions blindly before evaluating and comparing all steps and directions. The accumulation of small steps may cause the seeker to hit the wall, and this type of query should be augmented with "heavy" steps to bring the situation out of the way.

Formally, the main problem is to take minimal steps in the spaces of ambiguous codes and then direct these minimal steps through the spaces above to achieve the desired code. According to the teachings of this invention, the substantially smallest step in the space of ambiguous codes corresponds to a single pairing arrangement assigned to the code. At each step in the optimization method, the more pairs of pairs that can be tested are arranged better, if each pair of pairs is arranged optimally. Finally, the best match arrangement is selected to end all steps. If there is no pairing arrangement that improves the most in nature, one is randomly selected.

Referring to FIG. 7, the following steps are taken in the method:

□ 4000 selects a set of candidate codes as a start code.

□ 4001 uses perturbation method to obtain a new set of codes from the initial codes, and the permutation rule of symbol key-button pair is suitable, preferably the permutation combination of pair formula of all possibilities.

□ 4002 measures the attributes of the set of new codes.

□ 4003 checks whether termination criteria have been met, which refers to criteria that limit further improvement.

□ 4004 if the termination criteria has been met, the set of best codes is output.

□ 4005 if the termination criterion is not met, checking whether the code group obtained from the start code by the permutation method includes a group of codes better than the optimal code.

□ 4006 if there is a set of codes in the set of codes that is better than the best code, the set of best codes is selected as the new best code.

□ 4007 selects a new set of start codes. If the answer to step 4005 is yes, the best code in the set of codes is selected as the new start code, otherwise a new set of start codes is arbitrarily selected from the set of codes. After completion, return to step 4001.

When only one criterion is available for optimization, the selection of the best code from the set of candidate codes is performed by selecting the set of codes from the set of codes that has the best value on the criterion. However, when more than one criterion is available for optimization at the same time, the value groups of these criteria have only a local order, and it is not easy to choose among these value groups that is most favorable for the optimization process.

One way to do synchronous optimization is by performing optimization on each variable individually. Therefore, when the conflict occurs and the synchronization cannot be optimized, it is assumed that the general situation is used here, and it is necessary to establish the trade-off of the importance of each standard. This trade-off in relativity is part of the constraints imposed on the design.

The establishment of easy-touch codes to help explain how to make and use easy-touch codes, we will set three strict progressive criteria for easy-touch degrees.

□ A this degree of easy-touch typability is represented by an informal and yet acceptable typist characterized by 1) 20 words per minute and every 15 seconds of interference, i.e., on average every 5 words typist will encounter a query rate of questioning, 2) acceptable two percent of search errors, i.e., a search error rate of every 50 words or every two and one-half seconds.

□ degree B this degree of easy-touch typability is represented by a typist who is more acceptable than formal ones, and is characterized by 1) 20 words per minute and every 30 seconds of interference, i.e., on average every 10 words typist will encounter a query rate of questioning, 2) an acceptable percentage of search errors, i.e., a search error rate of every 100 words, or every 5 minutes.

□ degree C this degree of easy-touch typability is represented by a skilled typist that is characterized by 1) 40 words per minute and every 30 seconds of interference, i.e., on average every 20 words typist will encounter a query rate of questioning, 2) acceptable one-half percent of search errors, i.e., every 200 words of search, or every 5 minutes.

In fig. 8, we point out that the method for creating the easy-touch key code comprises the following steps:

□ 5000 determines acceptable quantization values for the lookup error rate and the lookup rate.

□ 5001A method for ambiguous code optimization is selected.

□ 5002 determines the minimum number of required keys, i.e. the number of keys and steps determined are used.

□ 5001 the optimization method selected in 5001 achieves the lookup error rate and the query rate determined in step 5000.

□ 5003 the maximum number of buttons allowed is determined in known typing device designs.

□ 5004 determines whether these design criteria are compatible. If the number of keys determined in step 5003 is greater than or equal to the number of keys determined in step 5002, the design criteria are compatible, otherwise they are not.

□ 5005 if the design criteria are compatible, the selected optimization procedure is applied in step 5001 to create an appropriate touch-typable ambiguous code, as determined in step 5004. This procedure is unsuccessful if not compatible.

The details of this method are as follows:

step 5000, determining the acceptable search error rate and the number of query rate. This can be done by testing a single or a group of typists, or directly by pre-selecting the values of the approved lookup error rate and query rate, for example, by selecting one of the criteria for the degree of typability according to the method described above.

Step 5001, a method for ambiguous code optimization is selected. With reference to the above method for establishing the optimal ambiguous code, two optimization methods have been discussed: random search and directed random advance. Random searching is not as efficient as directed random advancement, but may be sufficient if the number of allowed keys is large enough and the recognized easy-touch key count is low enough. Another weaker approach is to select a code in a single random trial, which may be more than adequate in some cases. To see this in more quantitative detail, please refer to the experimental results discussed in fig. 9, 10 and 11.

In this experiment, 5000 ambiguous codes and a distribution as even as possible were randomly selected from each set of ambiguous codes of 2 to 20 keys. Meanwhile, an optimization procedure is performed on each of the 2 to 20 buttons by using a guided random walk method, which is performed under three conditions of 1) optimizing only the search error rate, 2) optimizing only the query rate, and 3) optimizing both the search error rate and the query rate by using a prescribed numerical method. From the calculated values of the lookup error rate and the lookup rate for the randomly selected code, the following statistics can be derived: best, worst, average, and median. All the above statistics are shown in FIG. 9 showing the search error rate and in FIG. 10 showing the query rate, and the results of the optimization procedure are applied at the same time, in which the search error rate and the query rate are respectively the single ergonomic criteria for optimization. The results of the optimization process for optimizing both the search error rate and the query rate are recorded in FIG. 11. From these data, we can decide which optimization procedure to use. The most ergonomic approach is perhaps the most obvious choice, but sometimes a less standard approach is more than enough, e.g., any single random choice code will meet the specified criteria if they are loose enough. This will be discussed further below.

In step 5002, the minimum required number of buttons is determined, i.e., the lookup error rate and the query rate determined in step 5000 can be achieved by using the determined number of buttons and the optimized method selected in step 5001.

Referring to the above experimental results, and the selected easy-touch typing level, we can build a table to show the minimum number of buttons required for the three easy-touch typing levels, and refer to the above three optimized methods. It is such a table that is illustrated in fig. 12.

In step 5003, the maximum number of allowed buttons is determined in a conventional typing device design. Ambiguous codes will be most commonly used on small devices, and the number of keys is generally limited by the size of the keys and the size of the entire typing device. In some cases, a convention may affect the number of keys, such as the convention of using 12 keys on a telephone keypad.

At step 5004, a determination is made as to whether the design criteria are compatible. If the number of keys determined in step 5003 is greater than or equal to the number of keys determined in step 5002, the design criteria are compatible, otherwise they are not.

The number of keys that can be allowed on a typing device may be limited by a number of factors and these factors determine the degree of limitation, as will become more apparent below describing in detail the application of the device.

If the design criteria are compatible, step 5005, the selected optimization procedure is applied to step 5001 to create a suitable easy-touch typing divergent code, as determined in step 5004. If not, the procedure fails.

If such a procedure fails, at least one of the following situations may occur:

□ a more ergonomic optimization method is selected.

□ the design of the device is modified to accommodate a greater number of keys.

□ may accept a lower level of easy-touch typing.

□ the device is discarded.

Smart card smart cards located on the 9 th through 16 th alphanumeric keys are devices that are comparable in size to credit cards and contain computer components like processors and memory. Smart cards are currently used in applications such as security management and banking, but there are many other possible applications. This application illustrates that it is feasible to mount a smart card on a touch-typable keyboard and thus greatly extends the range of applications for which the device can be used. As a simple example, in applications in security administration and banking, the user of a smart card must remember a string of password digits for use with the device. However, if a touch-key smart card is used, a longer code string in natural language, easy to remember, may replace a shorter numeric code, difficult to remember. Smartcard-like sized devices under the teachings of this invention may be applied to include personal digital assistants manufactured by Franklin Corporation and sold under the REX trademark.

Under current technology, the size of smart cards substantially limits the transmission of information keyed on the card by a keyboard through complex and energy-consuming communication components. Therefore, the application of the smart card is to be popularized by a low-cost device for performing ergonomic and rapid information transmission by using a general touch-type dialer and a general touch-type audio generator.

Most telephones have 12 key buttons, each having a respective associated touch key sound, i.e. pressing each respective key causes the telephone to sound a particular touch key sound. However, the dual tone complex frequency (DTMF) standard in the world provides 16 touch key sounds, and a dual tone complex frequency tone generator mounted on a general telephone set has a function of transmitting the entire 16 touch key sounds. By using these additional tones, the 16 keys each have a dedicated touch tone and can be used for encoding alphanumeric symbol sequences. The greater the number of keys, the less ambiguous the codes will be, without regard to other factors. It is therefore the case that this application will be generalized to actually use all 16 touch key sounds to encode the alphanumeric sequence. Thus, devices that transmit ambiguous encoded information can be produced using readily available and low cost components.

The application has the following further objects

□ provide a touch-key keyboard for a smart card sized device.

□ provides a method to simulate a set of code symbols that can actually be generated by a device and that are larger than the set of code symbols.

□ provides an example of the primary ergonomic criteria for finding error rates in a device.

□ provides a combination keyboard and video display design suitable for smart card sized devices.

□ provides a divergence elimination apparatus that can operate with very little computer memory.

□ provides a system having more than one operational discrimination elimination apparatus, each of which is adaptable to its own range of computer functionality. In this case, the first disambiguation device is used to provide feedback to the user at the sending end of the communication, while the second disambiguation device is used at the receiving end of the communication.

We will now discuss in detail how this application can achieve these objectives.

The easy-touch key keyboard smart card device is small in size, so that only a few key buttons of practical size can be mounted thereon. If a portion of the card is to be reserved for the image display, the portion that can be fitted with the key buttons is further reduced. The preferred compromise between the degree of touchable keying for the actual size of the key required and the large number of required keys to allow ambiguous codes is in the range of 9 to 16 keys. Two possible device designs with a key number within this range are shown in fig. 16 and 15. The arrangement and utility of the key buttons and their relationship to other components of the smart card are discussed in more detail below.

Synthesis of coded symbols using sequence coding and meaningless decoding referring to fig. 13, let us divide a set of decoded symbols into two sub-groups: 1) one as a central group containing symbols to be linked to coded symbols thus producing a one-to-one relationship with the actual typing device, 2) one as an auxiliary group containing symbols to be linked to coded symbols thus producing a many-to-one relationship with the actual typing device. (step 100) As such, the method for synthesizing a coded symbol further comprises the steps of:

101 establishes an initial and potentially ambiguous code associating subsets of the central group with code symbols that are associated with physical typing devices in a one-to-one relationship that can be generated by the typing devices that physically represent the code symbols.

102 find short sequences of coded symbols, wherein the impossible decoding of the sequence of coded symbols forms a meaningful decoding.

103 establishes a relationship between the secondary and ambiguous possible codes as secondary sets of decoded symbols and the short sequence of encoded symbols found in step 102.

For example, let us connect 16 code symbols with 16 dual tone complex frequency (DTMF) signals, so that the specified tone will actually represent the code symbols. These tones will be denoted by (0, 1, 2, 3, 4, 5, 6, 7, 8, 9, #, a, B, C, D). We will use the letters A-Z as the central group of symbols and connect them to the actually representative code symbols, via the following first set of ambiguous codes

(0，aw)，(1，bi)，(2，cx)，(3，d)，(4，ej)，(5，fo)，(6，g)，(7，hv)，(8，ky)，(9，1)，(*.mu)，(#，n)，(A，pz)，(B，qr)，(C，s)，(D，t)，

Where the first part of each pair is an encoded symbol and the second part of each pair is a decoded symbol concatenated with the encoded symbol. The decoded symbol of the auxiliary set will be a single code set containing the "space" symbol. From this we synthesize an encoded symbol representing the decoded symbol "space". The candidate sequence is A8A, which corresponds to the following coding sequence (pkp pkz zkp zkz pyp pyz zypz). None of these decoded sequences forms any of the characters listed on our sense sequence reference sheet, so the encoded sequence A8A is a part suitable for representing the helper set, and we form the pair of combinations (A8A, "space") to represent the symbol "space". That symbol "space" can be associated with a typing device that causes the audio sequence associated with A8A to be generated each time the designated typing device is activated. At the receiving end, a decoding device will convert the sequence A8A into "space" symbols. The known key input device is completely transparent to the user, whether it is connected to a single, actually representative coding device or a composite coding device. An arbitrarily large auxiliary set of decoded symbols can be represented. It can be seen that if there is a set of reference data and a first ambiguous code for a central set of code symbols, a programmer of ordinary skill will be able to readily create software that automatically generates any desired value for the composite code symbol.

A divergence elimination apparatus is provided that is capable of operating with a very small amount of computer memory. The limited processing capabilities and memory capacity of this generation of smart cards substantially adds to the ambiguity in the design of the ambiguous code, i.e., only a few of the computer functions on the card are dedicated to the ambiguity resolution device.

For any ambiguous code, most of the ambiguity resolution benefits, both in the computer hardware and software portion and in the user portion, occur in the selection of which other decoding should be used for the ambiguous code. Due to the limited calculator functionality of the smart card, the question of other decoding will be completely disregarded. Without considering the query, each code need only store the most probable decoding for it, since only the most probable decoding can be output via the disambiguation device when the respective code sequence is received. In brief, a particularly compact database, for example, in the form of a simple suffix tree, is available. Since the query need not be considered and sufficient visual feedback can be provided by using a simple and power-saving display, a single-line action-tag display associated with a display via a pocket computer or digital watch is an example.

This method of ambiguity elimination, which only stores and outputs the most probable decoding sequence, will be referred to as the easy-to-find ambiguity elimination method. Easy-to-find ambiguous code elimination operates effectively only with ambiguous codes that are sufficiently strongly touch sensitive. Thus we have concluded that a surprising conclusion can be drawn regarding codes with sufficient touch sensitivity, i.e. a strongly simplified disambiguation apparatus can be operated efficiently with sufficient touch sensitivity.

A standard 16-key functionally-perfected ambiguous code applied in this invention consists of a code with a lookup error rate of once per 4043 words and a query rate of once per 68 words, namely the code aw bi cx d ej fo g hv ky l mu n pz s t, as applied in the 16-letter key smart card standard design 51 shown in fig. 15. The figure also includes a display 50 designed to display encoded symbols entered via a keyboard and a thumb activated auxiliary entry device 51 that can be used for a number of additional symbols and mode conversions. This will be discussed in more detail in other applications. It is noted that the display 50 is preferably operable on the key input devices 51 and 52 if it can be positioned as follows: 1) the alphabet typing means 51 and the thumb activated auxiliary typing means 52 are placed in a comfortable position for single hand operation (right hand is the example in this figure) while allowing maximum button size, 2) and allowing a comfortable and full view of the display screen across the width between thumb and index finger, within the limits of the smart card size. Such unique and specific mounting solves the problem of how to effectively co-exist a touch-typable keyboard and a maximum possible display on a smart card.

Since the sender cannot allow the query in this application, this code is selected via a guided random forward optimization using the search error rate as the only criterion for optimization. It is noted here that the error rate of the search using this encoding will occur approximately once every 16 pages of typed text. This encoding is therefore suitable for the exact transmission of substantially lengthy information, albeit in the absence of querying means. An alternative code optimized for the query and lookup error rate criteria is aw bu cx d pv pz go hv im ky l NQ pr s t, provided that the sacrifice of the optimality of the partial lookup error rate at the receiving end is believed to help reduce the query processing at the transmitting end. The lookup error rate for this encoding is once every 2670 words and the query rate is once every 101 words. Selecting an encoding optimized according to the lookup error rate and the query rate will be appropriate in at least two cases, 1) if the smart card is sufficient to support the querying device, and/or 2) a query-based disambiguation device will be employed at the communication receiver activated by the smart card. For example, a user uses a smart card to make information, and then transmits the information to another computer via a telephone line, and after a period of time, uses a more powerful disambiguation apparatus to perform a second disambiguation. Indeed, the second disambiguation need not be performed by the original message producer, but may be performed by a second user, such as a secretary of the original message producer.

In any event, the second type of code lookup error rate is still very low, for example, a very skilled typist may still miss the key with a probability of about 1 per 100 words. In any reasonable case, both 16-letter key codes need to be easy-touch, so that a typist using the first 16-letter key code will only answer a query every three minutes while typing 20 words per minute, while using the second 16-letter key code will only answer a query every five minutes. When the same optimization procedure is carried out on 9-letter key type ambiguous codes, the group of codes akw by cly dhx epv fim gr jot suz is found to carry out the optimization procedure only on the search error rate, wherein the search error rate and the query rate are respectively once every 116 words and once every 4.4 words. This set of codes has been used for the 9-letter key smart card standard design shown in fig. 16. To optimize the search error rate and the query rate by using the guided random walk method, we can simultaneously establish a set of codes with search error rate and query rate respectively being every 109 words and every 6.2 words, for example, a set of codes of am bnz cfi dhx evw gjr kosluy pqt. It is noted that since queries cannot be performed using easy-to-find ambiguity resolution, the degree of easy-to-touch can only be discussed by referring to the search error rate. Here we need an evaluation as to whether the lookup error rate is low enough to make an acceptable word. Despite the 9-letter key codes, the search error rate is comparable to the probability of a skilled typist pressing the wrong key, and thus these codes can be considered as easy-touch codes under this premise. Further, since smart cards are most commonly used for transmitting short messages, making e-mails, communicating between pagers, etc., for example, the level of text correctness may be lower than the level of transcription of the final text. From these considerations we have derived the definition of 9 to 16 keys as the preferred number range of keys for this application. If there are more than 16 keys, it is not easy to retain the advantage of a substantial key size while mounting on a smart card. Conversely, if the easy-to-find disambiguation device is compatible with the limited computer functionality of the smart card, then a differential code on less than the 9-letter key may be non-touch-typable.

A smart card with an easily searchable ambiguity resolution device for typist feedback can be operated by a simple touch typist by voice synthesis over telephone lines based on symbols typed by typists on smart cards without feedback on the progress of the communication from the card and/or from the communication sink device. However, it is also feasible to provide feedback directly from the smart card, with sufficient computer resources to provide the smart card support feedback.

It is noted that less computer resources are required to provide effective feedback than are required to provide feedback from an easy-to-find disambiguation device. Although there is no software associated with the disambiguating database on the smart card and only the electronic circuit board of the original shape is used, the skilled user knows that a particular font is directly transferred to the display when the corresponding key button is pressed, and that the font is the letter most likely to be associated with the key button. Taking the above-mentioned group of codes aw bi cx d ej fo g hv ky l mu pz qr s t as an example, the completed text is usually readable and understandable for human beings. Using the 1-block (single letter) data, taking the first row of the Gettysburg address as an example, the following can be read:

oour score and sehen kears ago our oathers irought oorth onthis continent，a nea nation，conceihed in liiertk，and dedicatedto the proposition that all uen are created erual.

such a level of accuracy is sufficient to provide the typist with a rough guideline as to the text he or she is typing on the smart card. This example illustrates that discrimination elimination can be achieved with an extremely small amount of memory; the only memory required here is to store the 16 fonts to be displayed correspondingly when the 16 keys are activated. This process is scalable with the required computer resources. The more memory, e.g., 2-, 3-, the higher the probability of a block can be stored and used as the basis for the famous block-based disambiguation procedure, thereby allowing the user to display text modifications with increased accuracy.

Although block-based disambiguation procedures are well known in the art, they have so far proven impractical. The following example will illustrate this reason: the block-based disambiguation procedure is not efficient enough to discriminate too ambiguous codes. The block-based disambiguation program of this example incorporates a code that is sufficiently touch sensitive and sufficiently ambiguous so that the efficient disambiguation program can proceed block-based. The conventional method has been to use a block as a main ambiguity elimination program and a character as a main ambiguity elimination program. However, applying the teachings of the present invention will make block-based disambiguation procedures operable and practical.

This example further illustrates: 1) no character-based disambiguation procedure is required when using the teachings of the present invention, 2) no microprocessor is required when using the teachings of the present invention, 3) more than one and possibly different disambiguation devices can be used for the same communication system based on the ambiguous code. The character-based disambiguation procedure, or another disambiguation method, may be used at the communication receiving end of the smart card that is sent from the user using a simple block-based disambiguation method to provide feedback to the user of the smart card.

When the block length of the block-based ambiguity elimination method is increased, the character modification accuracy is increased. However, the memory capacity required to reach certain block lengths is gradually approaching that of the character-based disambiguation procedure, and the character-based disambiguation procedure will usually choose to use it if there is sufficient memory to support it, because it generally yields better results than the block-based disambiguation procedure.

Further applications the potential applications of the device can be greatly increased if there is available memory in the smart card beyond what is needed to store the disambiguation database and software. For example, if there are a few more memories, a user can query the phone book for information via a suitable voice messaging system, can type a name and other similar attribute information on the smart card keypad to query a phone number, and can store the queried phone number in the user's memory for later download to another better performing device.

Query minimization-typing devices designed for vehicles this application takes into account a situation when queries are the main limitation in the design of typing devices. Generally, it is advantageous to reduce the query rate, which reduces the need to answer queries during typing. In some applications, however, reducing the query rate is of high importance.

The query will be displayed on an image display for the most practical application of typing devices using ambiguous codes. In situations where the user is in urgent need of vision, such as when the user is driving a vehicle, safety considerations may force the visual impact of the query evaluation to be minimized. Although the query is made via an auditory device, it is extremely important to minimize the distracting factor of the driver. In addition, when driving a vehicle, it is often necessary to hold the steering wheel with both hands, preferably without removing the steering wheel by requiring operation of a typing device. This object is achieved by a typing device for mounting the typing device directly on the steering wheel.

Referring to fig. 17, we will find that any keying device may be mounted on steering wheel 200. Many steering wheels have a pinch-grip attached to the inner and outer skin layers to enhance finger grip. For such a steering wheel, it is of course the case that the first plurality of keying means 201 is associated with a respective plurality of these pinches. When the driver grasps the steering wheel, the fingers of each hand will touch four of the first key input devices 201. The location where the driver touches the steering wheel may change from time to time, for example when the driver turns the steering wheel at a large angle. Which 8 key groups the driver touches at any one time can be confirmed by a position sensing means, such as a combination of a pressure sensing means and a simple electronic circuit board, which combination will be obvious to a skilled user.

The second plurality of key input devices 202 may be mounted on the steering wheel on the inner or outer layers, and the thumb of each hand will touch a designated one of the second key input devices while the driver is able to hold the steering wheel. Which second key input device the driver's thumb will touch at any one time can be confirmed via a suitable position sensing device.

As for the above-described keyboard mounted on the steering wheel, it is natural that one code is selected among 8 buttons, and the known buttons are connected to the first key input means touched by the fingers of both hands, and the two mode-switching buttons are connected to the second key input means touched by the thumb of the driver.

Ambiguous code selection the guided random forward method taught by the present invention is applied to the selection of substantially optimal codes, and we can establish the following codes on the set of 8-bit letter keys, based only on the query rate as the criterion for optimization: aksz bcev dfi gmo hgt jnw lux prx. The lookup error rate and the lookup rate are once every 70.2 words and once every 4.1 words, respectively. As such, these ratios are calculated based on our reference data using a simple character-based disambiguation procedure as a method of disambiguating. The query rate may be too high if the code is to be considered touch-sensitive. A 20-word-per-minute typist or driver may be distracted from driving by a query approximately every 12 seconds, which may be overly frequent if necessary to be compatible with safe driving regulations. On the other hand, a skilled typist may not be able to drive a car while typing 20 words per minute, and thus may bring the relationship between query and typing speed into an acceptable range for easy-touch typing.

Several additional strategies to reduce the query rate have gone beyond the selection of a substantially optimal code and these strategies can be used in combination. Comprises that

□ the total number of key buttons is increased by increasing the number of key buttons that can be activated by a single finger. This may be advantageous, for example, by adding a row of buttons to the steering wheel, or by equally positioning each button in multiple positions, or by using a coordinated approach, i.e. by encoding different subsets of code symbols when more than two buttons are pressed simultaneously.

□ eliminate the query when the likelihood of the low likelihood decoding process differs too much from the likelihood of the most likely decoding process. The parameter that controls how close the probability between the low probability decoding process and the most likely decoding process should be to the most likely decoding process must trigger a query, and the parameter value can be selected by the user. Such a mechanism would be of value in any application where the query rate is a relevant ergonomic criterion.

□ use a method that combines coordinated/ambiguous codes, as described in more detail below, with a more powerful disambiguation method than the simplified character-based disambiguation method.

Keypad compatible with existing telephone keypad

In this application, the limitation of the number of keys is significant in the case where the keypad needs to be compatible with existing telephone devices, which typically have 12 keys. In this application, it is desirable to retain two keys for non-alphabetic symbols, such as the space key, the elimination key, the period key, and the end of transmission key. Therefore, the 26 letters must be distributed over a maximum of 10 keys. In this application, a minimum lookup error rate and a minimum query rate are also required. We have found that using the optimal method and 10 keys equally spaced as much as possible, we can find a set of codes with a lookup error rate of 138 words per time and a query rate of 9.3 words per time, such as amq be cdu fiygqx hl jsv krz nw ot, and optimize both the lookup error rate and the query rate. This is an overall improvement of over 4 times over standard ambiguous codes, compared to a lookup error rate of 29 words and a query rate of 2.2 words. In fig. 18, the application of 10-key coding to the standard design of existing phone keypads with optimized lookup error rate and query rate is illustrated.

We can also compare this 10-key code with the 9-key code proposed in the U.S. CITE tegic patent and the EPO patent of CITE EPO application. The first group of these codes, afg bkn jlo mqr dhi sux ptv cyz, has a lookup error rate of 86.5 words per time and a query rate of 3.9 words per time, the second group, rpq adf nbz olx ewv img cykthj su, has a lookup error rate of 115 words per time and a query rate of 5.2 words per time. These codes are all significantly inferior to the 10-key code designed for this task. While neither the U.S. patent CITE tegic nor the EPO patent application CITE EPO were generated for the creation of such an ambiguous code, and these data did not exist when the code was optimized (if indeed optimized), we were unable to make a decision on the substantial optimality of the code.

Another useful comparison is to a 9-key code optimized for lookup error rate and query rate. For example, we establish a set of am bnz cfi dhx gjr kos lue pqt with a lookup error rate of 109 words per time and a query rate of 6.2 words per time. Comparing these results, we have found that the improvements obtained by this applied guideline are two issues 1) using more than 9 buttons so that the search error rate and query rate can be improved, and 2) optimizing both the search error rate and query rate. Several approaches described in this application will expand to the application of 11-key coding and 12-key coding, and we will find that the search error rate and query rate of a set of 11-key codes, such as avy bn cl dhx ewfip gjo kr mu qt sz, are 215 words and 10.1 words, respectively, while the search error rate and query rate of a set of 12-key codes, such as aw bn cky dhgq ef go ip jr lz mxsv tu, are 313 words and 13.2 words, respectively.

Thus, by sacrificing the use of the # key and the # key when encoding non-alphabetic symbols, we can significantly improve the search error rate and substantially improve the query rate, thereby easily bringing a standard telephone compatible keyboard into (degree B) the easy-touch range. Whether these improvements can compensate for the inability to use the # key in non-alphanumeric coding and the loss of the # key can only be determined by the intended function of the device being built. It is noted here that the non-alphabetic symbols can be encoded in the code sequence of the smart card application described in the foregoing. In the event that non-alphabetic symbols can be encoded using the x and # buttons, a particularly ergonomic design for delivering end symbols using the x and # buttons is followed in part as follows. Let # be a blank symbol ═ character enter stop symbol code, ## be a sentence enter stop symbol code, and # # # be a transfer stop symbol code. The complexity of such a symbol encoding is inversely proportional to the probability of occurrence of the symbol. Depending on the application, the sequence of symbols can be used to encode other non-alphabetic symbols, such as the dismiss key, @ (which is the case in e-mail applications), and/or as mode-switching symbols.

The use of telephone keypad with alphabetic English letters provides a solution to the severely limited keyboard design, which refers to a keyboard with fixed number of keys, fixed positions of keys and fixed arrangement of symbols on the keys. This problem arises in keyboard design, such as 1) preserving as much as possible the alphabetical order that is ultimately used by standard ambiguous codes, 2) being compatible with existing standard telephone keypads, and 3) having improved search error and query rates compared to standard ambiguous codes. These constraints limit the freedom in selecting a number of keys that will be the basis for ambiguous codes. For example, if one letter is selected to occupy 10 keys on the keypad of a telephone, the alphakey and the # key can be used to encode non-letter symbols. Also, in the case where the standard ambiguous codes are used with as equal a separation as possible, we can choose an alternative separation while still following the known constraints.

Given the limitations of alphabetic alignment, the individual alignment of the 26 cells is separated into 10 groups associated with a unique ambiguous code. If the processing time of the calculator is sufficient, it is possible to evaluate the respective lookup error rate and query rate of the codes. A different and more efficient procedure would be to apply the optimization method proposed by this invention to this optimization-limited problem. The invention proposes that in the absence of information suggesting the use of some more complex basic steps, it is necessary to define a most simplified basic step for the possible code groups. In the present case, the ambiguous code is a list of 10 letters, all of which are included in a group, and the letters appear in alphabetical order. An example is ab cd efgh ij kl mn opqr stuv wxyz. So that there are 9 spaces in the group for the separation. One basic step is to move a letter across a space. For example, if we select the second space, we can get the abc d ef gh ij kl mn opqr stuv wxyz code by shifting the letter c to the left, or the a bcd ef gh ij kl opqr stuv wxyz code by shifting the letter b to the right, in one basic shift. When this particular code is known, all possible codes that can be reached by one elementary shift from this particular code can easily be generated. Given this finding and given the guided random advance described above, it will be apparent to a skilled user how to apply the optimization method taught by the present invention in the present situation. For example, applying this method we find the set of ab cd ef gh ijklm no pqr s tu vwjyz with a lookup error rate of 65 words and a query rate of 5.8 words. This code is ideally arranged on the keypad of the telephone as shown in figure 19. The error rate of this code should be compared to the standard ambiguous code with a lookup error rate of 29 words and a query rate of 2.2 words. After comparison, it was found that the search error rate was improved by more than 2 times, while the query rate was improved by nearly more than 3 times, without affecting the alphabetical order and compatibility with existing telephone devices. It is worth mentioning that the above discussion of the selection of 11-key or 12-key alphanumeric codes for a telephone device is also applicable to this application; substantially optimal codes for 11 and 12 keys can be obtained using partition optimization.

This compartmentalization optimization method is obviously not only applicable to this application; for example, it can be applied to the smart card device discussed above, thereby obtaining an optimal code in the alphabetical arrangement of 9-16 key buttons with alphabet symbols in a row.

The Qwerty-like keyboard is used in the previous application to create a keyboard 1) that is immediately compatible with standard keyboards and 2) that is immediately optimized for different ergonomic standards, and can be used to create a keyboard 1) that is similar to standard Qwerty keyboards and 2) that is optimized for different ergonomic standards. As in the previous application, we will maintain the design of the standard keyboard by keeping the symbol-to-key assignment arrangement as much as possible, and simultaneously optimize the separation of these arranged symbols, thus also minimizing the search error rate and query rate. Such applications require a further limitation in that the letters are left in the same column of key buttons as in the known Qwerty arrangement.

There is a Qwerty-like keyboard design sequence, that is, three columns dedicated to alphabetic buttons, and a different number of columns, perhaps one to 10 columns. Obviously, with only one column, i.e., three buttons, the search error rate and query rate will be extremely high, and only one possible ambiguous code corresponds to the arrangement of the Qwerty keyboard symbols. This code is known as qwertyuoop asdfghjkl zxcvbnm, with a lookup error rate of 2.8 words once and a query rate of 1.1 words once, and a code of such poor quality is unlikely to be accepted as human and practical. With the increase of the number of columns, we will be able to find better and better ambiguous codes. Meanwhile, as the number of columns increases, the size of the key code required to be installed in the device remains substantially full-size key codes, and increases. Thus, a Qwerty-like keyboard must be a compromise between coding diversity and keyboard size. For example, if we wish to make a keyboard that is similar to the Qwerty style and the same size as a pocket calculator, but uses full-size key code, then 7 columns may be used, as shown in fig. 20. A substantially optimal touch-typable code based on lookup error rate and query rate would be qwe r t yu I o asd f g hjk l zxc vb n m with a lookup error rate of 668 words per query rate of 35.5 words per query rate, which is apparent for many more or less typists and keyboard applications. In fig. 20, this encoding is illustrated in an ideal arrangement. The keyboard-attached typing device shown in this figure is suitable for use in taking notes, writing e-mail, and the like. It will be ready for typing and will require no or minimal learning by anyone familiar with the standard Qwerty keyboard, and can be easily placed in a pocket even though it is made with full-size key codes.

It should be noted that depending on the lookup error rate and the lookup rate, the cost of attaching to the conventional is quite high, although only approximately attaching to the conventional. If we can now have 17 additional assignments of letters to keys, we find codes such as w r t bu gi ov p af d ej ky l hz cx n mq with a lookup error rate of 7483 words once and a lookup rate of 290 words once. This corresponds to one error per 30 pages of the typewritten document and less than one query per page of the typewritten document. Such a device would be difficult to imagine if not of the touch-key type.

Referring to fig. 21, we will see that this code can be designed with 18 letters at or very close to their Qwerty position, which will be indicated in bold. In this arrangement, to maximize the similarity of typing between Qwerty keyboard typing and the optimized Qwerty keyboard typing, the fingers should be placed in the original column, i.e., left index finger on (space bar) and right index finger on (ej key). It can be seen that by simultaneously keying the space key with 'e' on the original column, the design will step forward from the Qwerty design in terms of structural accuracy, and the dependency on the most flexible finger will be relatively higher than that of the Qwerty design. By proper arrangement of the keybutton symbols, any ambiguous code can be optimally matched to a Qwerty (or other conventional) keyboard.

It is noted here that the tolerance is slightly different from the strict Qwerty arrangement, which results in a substantial improvement in the utility of the Qwerty similarity. When the rows differ from each other only slightly, as in the standard Qwerty design, many or most of the finger movements required to operate the Qwerty-like keyboard are the same or similar to operating the standard Qwerty keyboard. The above describes a conditional exchange between ergonomic criteria for preserving legacy order and ergonomic criteria for preserving legacy functionality.

In view of the variety of criteria that must be optimized and the different needs of the user, it is desirable to have the facility practical to select between the optimized qwerty keyboard and other keyboards that are optimized for ergonomic criteria in terms of search error rate and query rate. The ability to change the key tag within the software facilitates the decision of the above selection. To achieve this, the keys need to have the ability to display more than one symbol at a time. The display function can be formed by a light emitting diode or a liquid crystal display screen.

Skilled users will find that modern keyboard design methods can be applied to the retention and partial retention of other conventional keyboard designs, such as the azerty keyboard used in france.

The purpose of applications such as the numeric keypad is to allow most computer users to enjoy the advantages of the ambiguous keyboard while maintaining the same hardware and at the lowest cost. These clearly include the advantages of one-handed typing and compatibility with palm-type compatible discriminative keyboards. Standard 101-key keyboards for workstations and personal computers typically include a set of numeric keys on the right end of the keyboard in a qwerty configuration. There are usually directional keys or functions that can be used to move a cursor near the set of numeric keys.

FIG. 22 represents the feasibility of integrating the ambiguous code with the function of moving the cursor 601 optimized for the conventional numeric key configuration 600. The numeric key arrangement 600 described above has 17 keys in this example of different sizes. Depending on other design constraints, all or some of the above keys may be used for punctuation or other symbols. These design constraints may affect the choice of the number of buttons assigned to the letters, the distribution of the letters and other symbols in different patterns, etc. The main features of this application include:

□ assigning ambiguous codes to a plurality of keys of a number keypad

□ replacement mode usage with a secondary thumb-activated key-in assist function

Skilled users will find that the ambiguous code assignments for the numeric keypad described above can be made by software; no special function hardware is necessary. If the assigned key is to include the property of ambiguous codes, the key label must be changed. For an example of an entity that uses ambiguous codes in this setting, the ambiguous codes assigned to the 17 buttons must be selected based on the existing standard corpus and on the premise of lookup errors and query rates. The code af bu cx d ej gi hz ky l mq n ov p r s t w shown in fig. 22 has a lookup error rate of one lookup error per 7483 words and a query rate of one query per 290 words. This encoding has been discussed previously. The arrangement of the codes is kept as same as the English letter sequence as possible. This known code is not optimized in alphabetical order; but only optimize the search error rate and the query rate. In accordance with the teachings of the present invention, the search error rate, the search rate, the alphabetic order, and/or other ergonomic criteria may be simultaneously optimized.

FIG. 22 may be used to illustrate an alternate mode usage where the thumb-activated assisted keying function is secondary. We can assume from this figure that there are 4 keys that constitute the auxiliary key function: up 602, down 603, left 604, right 605, etc. directional keys. These functions are usually built on 4 depressible keys, but sometimes on a keypad, a rocker, or any device that can generate a plurality and different signals under user-operated conditions.

It is noted in FIG. 22 that the plurality of keys are labeled with different symbols than in the ambiguous code, which in this example are all numerical codes. These symbols are obtained by depressing a designated one of the 4 additional key functions. The typing functionality on the auxiliary keypad can be assigned as a mode as follows:

□ 602 move the key (up) to get capital letters.

□ 603 (down) number/punctuation pattern.

□ 604 the (left) key button is marked with a left symbol.

□ 605 the (right) button is marked with a right symbol.

It is worth mentioning that 1) according to the teachings of the present invention, we can use other types of symbol and/or pattern assignments on the auxiliary keypad, and 2) more symbols and patterns can be assigned to more complex auxiliary key function modules. We will discuss the mode assignment of symbols in more depth according to another application. This discussion will apply to this and other applications.

The objectives and advantages of 13-letter key-type coding many related applications in the teachings of the present invention benefit from the surprising advantages of ambiguous code optimization; this is that the number of key buttons dedicated to extremely relevant symbols is half the number of extremely relevant symbols. If we choose [ a-z ] of the English alphabet as the extremely relevant symbol, the ideal number of key buttons is 13. The 13-letter key type coding in english includes the following surprising advantages:

□ the key-touch degree is easy to touch,

□ is ergonomic, can be typed by touching keys, is not a ambiguous character typing method,

□ is ergonomic, can be typed by touching keys, and can be used for inquiry,

□ compatibility with standard keyboard configurations (qwerty keyboards, numeric keypads, telephone keypads),

□ provide for the preservation of one-handed to two-handed typing techniques,

□ provides a mouse/keyboard complex,

□ provides a means to reduce typing injuries.

Further objects and advantages will appear from the following detailed specification.

Easy-to-touch degree it can be seen from fig. 11 and 12 that even a skilled typist considers that the ambiguous code on 13 keys has a high easy-to-touch degree in the character-based ambiguous code elimination method. Using the above-described guided random walk, we find that the code aw bn ck ef go hgip jsly ly mx qtrz has a lookup error rate of one lookup error per 515 words and a lookup rate of one lookup per 21 words, which is c-level easy touch. Fig. 25 shows an ideal arrangement of this code. As previously mentioned, changing the variables that control the query importance and whether the user is attentive can further reduce the query rate. In the range where the query is not used at all, the degree of easy touch is controlled by the search error rate. With the above encoding, there is about one search error in the text entered every two pages on average. This ratio is far less than the error rate of a skilled typist. The 13-key button code is then suitable for use in a variety of different touch typing jobs and users.

It is very convenient to use any ambiguous keyboard and any ambiguous code elimination method, such as the numeric entry of an ambiguous elimination database, in a non-ambiguous information entry manner that conforms to ergonomics and allows touch-typable non-ambiguous character entry. Among all the uses of the ambiguous keyboard, the more ergonomic the non-ambiguous character typing method is, i.e., the simpler the operation is, the better the operation is. Ambiguous keyboards are essentially for touch typing, and it is more convenient to operate such keyboards in touch typing as well as non-ambiguous information typing.

To achieve non-ambiguous text entry using a small number of buttons, a common strategy is to use a coordination method. To meet the need for using the minimum number of keys, the coordinator designers are constantly researching forms that simplify the coordinator process with a sufficient number of keys. It is basically proven that the complexity of the coordination method cannot be more than 2, i.e. it is not necessary to activate more than 2 key-in functions simultaneously to achieve non-ambiguous symbol codes, and the number of keys cannot be less than half of the number of symbols to be coded. The present invention is in contrast to the above-described techniques and indicates that the number of key buttons cannot be less than half the number of symbols to be encoded if the simple function of non-ambiguous symbol encoding exists. In a more specific example, the present invention indicates that at least 13 buttons are required to represent the letters [ a-z ], and at least one mode-switching button. When the mode-switching key is used simultaneously with other keys, the letters associated with the above keys can be uniquely and indiscriminately encoded.

On a divergent keyboard, some of the keys need only be pressed once to represent a number of symbols. If the same keyboard is to be used in the non-discriminative mode, the single key button must be combined with at least one other key button, perhaps using the key, to select each symbol associated with that key button. The more simple and better the key button combination is in terms of ergonomics. For easy key touch, the same key button combination would be ideal for non-ambiguous typing of all symbols. To achieve both of these criteria, 1) the number of symbols assigned to each ambiguous key must be the same, and 2) the number of symbols on each ambiguous key must be reduced. In general, the number of these ideal standard keys for ambiguous codes would be half the total number of symbols on the ambiguous keys. For example, the above standard means 13 is the ideal number of buttons representing 26 English alphabet symbols.

The above findings can be referred to in fig. 23 and 13-key ambiguous representation of english letters, and further studies can be made on the use of an ergonomic and touch typable ambiguous keyboard using an ergonomic and touch typable non-ambiguous character typing mode. As can be seen in the above figures, each of the divergent representative letters 700 encodes only two letters in a subset of the representative letters 700. The touch typable keyboard also includes a mode button 701 and functions to translate ambiguous and non-ambiguous typing methods. The functionality of the ambiguous and non-ambiguous typing methods may be controlled by software to switch modes as required by the circumstances, or to assign a certain key button according to the switching mode, or a special type of typing functionality, such as mode switch key button 701 using dual keys. In the non-ambiguous typing mode, if the buttons of 700 are activated at approximately the same time as the 701 buttons, the 701 buttons are activated to encode one of the two symbols associated with the 700 buttons.

It is desirable to describe the symbol pairs of the 700 keys as left and right symbols, with the left symbol being marked to the left of the key and the right symbol being marked to the right of the key. Without loss of generality, to achieve non-ambiguous text entry, the left symbol to the left of the keybutton is associated with 701 keybutton activations. When any of the 700 key sets and the 701 key are simultaneously activated, the left symbol is selected unambiguously. If the same key in the set of 700 keys is not simultaneously activated with the 701 key, the right symbol is selected unambiguously. The non-ambiguous text entry method described above is also applicable to keyboards that incorporate other modes.

The method is suitable for touch typing type inquiry method, even by using optimized ambiguous codes, unlimited computer processing capacity and unexplored artificial intelligence ambiguous elimination technology, ambiguous sequences can still be generated when characters are typed, and the optimal effect of ambiguous elimination can be achieved by the intervention of a user.

A true touch typist can bet on the typed text or the original text without looking at the keyboard. For touch typists, it would be desirable to arrange to set all the interpretation of ambiguous sequence queries to 1) not to affect the typist's attention to the screen, and 2) to be able to reply to all queries from the keyboard in a simple and fixed manner. To achieve the above objective, a series of candidate characters can be accessed and ambiguous characters selected on the screen using a reserved key. The user can operate the scroll bar button to scan the candidate characters by character, and the characters in the scan field are presented in a selected state in the case where the scroll bar button is pressed in addition to the character.

Reference is made to FIGS. 23 and 24 for a more detailed discussion of the software underlying the adaptive touch typable query method and the visual display controlled by the software. A query is detected in the first step 800, i.e., the disambiguation function finds more than one meaningful set of decoded sequences in the database corresponding to the entered encoded sequence. The function after entering the query mode will divert the user's attention to the displayed query interpretation. Such functionality may be visual functionality like the box 702 in FIG. 23. The possible decodings are then arranged in order of their likelihood (step 802). The highest likelihood decoding is then displayed (step 804) in the attention function area on the screen as described above. The software waits to receive a scroll bar button or other button entry in the ready state (step 806). If other buttons are being entered, the function of diverting the user's attention is removed (step 808), and the code is added to the previously entered text (step 810) and the method returns to ambiguous text entry. Conversely, if at step 806 a scroll bar button entry is detected, the database 812 is used as a reference to test whether meaningful interpretation is obtained. If so, the existing decode will be replaced with the next more likely decode (step 814) and then go back to step 806. If there is no more likely translation 9 the typing device enters the non-ambiguous text-entry mode (step 816), the translation sequence enters non-discriminately, the function of diverting the user's attention is removed (step 808), and the code is added to the previously entered text (step 810) and the ambiguous text-entry method is returned (step 818).

Although the above approach representing other options is described in terms of adapting a touch typable query method, the same approach may be applied in other situations. For example, when "code" represents a character with associated meaning, the database is a thesaurus, or when "code" represents a character translated into various meanings in foreign languages, the possibility of coding is provided by automatic translation software.

The cross-platform design is reserved to meet ergonomic standards: mouse/keyboard assume that a user of a palmtop device, such as a personal digital assistant, having a built-in one-handed typing keyboard, usually takes one day, and is likely to also use the two-handed keyboard of a calculator. If the user wants to effectively use touch typing on both devices, the more similar the movement patterns used on one-handed and two-handed keyboards must be. The shorter the time to switch between using the two keyboards, the more important it is to preserve the typing skills. The invention provides a device for quickly switching between a single-hand keyboard and a double-hand keyboard, and can achieve the reservation of cross-platform typing technology.

The invention relates to a single-handed keyboard suitable for typing tabular data for use in a spreadsheet or network. This invention can usefully interact with programs such as video games or graphics programs, 1) requiring rapid cursor and typing replacement, and/or 2) when the suitability of the symbol for typing depends on the position of the cursor on the screen. When using the qwerty keyboard and mouse with a standard calculator, the user must remove his hand from the keyboard to operate the mouse. The interaction between the mouse and the keyboard is very slow and cumbersome in work involving rapid and continuous typing and mouse operations, such as marking a design drawing on a calculator screen or filling in a form such as a web page. In this invention, the one-hand keyboard is mounted on a structure capable of sliding on a plane, so that the mouse and keyboard functions can be achieved. Although it would be preferable for the user to consider it to be primarily typing or to use a two-handed keyboard, the more similar key button arrangement between the two keyboards would be more desirable when used in combination between a single-handed and a two-handed keyboard, providing seamless transfer of typing techniques between the single-handed and the two-handed keyboards.

It can be seen from figures 25, 28 and 26 that the ambiguous code that can be arranged on a one-handed keyboard is selected such that the finger and thumb motions (in the case of the right hand in the figure) are the same on either a one-handed or two-handed keyboard, and the typing motion of the other hand is similar to the one-handed typing that was specifically selected.

This design strategy is as follows:

□ select a 13-key button code with sufficiently low search error rate and query rate.

□ is the appearance of the 13 key selection arrangement described above. The ideal arrangement would be 5 keys in the top row, 5 keys in the middle row (reference row), and 3 keys in the bottom row.

□ selects which hand to use to activate the one-handed keyboard.

□ the key buttons are arranged according to the previous step of left-right hand selection. And:

-using the specific gravity of the reference row as maximum.

The specific gravity of the most common finger is at a maximum.

The specific gravity of the uppermost row is higher than that of the lowermost row.

Then, all the left-hand activated keys are paired with the right-hand activated keys to obtain the arrangement of the two-hand keyboard, and the left-right paired keys must be symmetrical to a plane with a straight line drawn from the center of the keyboard from bottom to top.

Selecting one of the two symbols assigned to the original 13 keys, and associating the selected symbol with one of the keys paired in the above step. This step may be performed in a manner that favors a one-handed keyboard or an associated two-handed keyboard.

Biased on single-handed keyboards: the keys with high use degree are arranged on the selected side of the double-hand keyboard, and the keys with low use degree are arranged on the other side of the keyboard.

Biased on the two-handed keyboard: the keys with high or low use degree are arranged at the same side of the original 13 keys on the double-hand keyboard, so that the keys are kept at

The total number of possibilities for the degree of use on both sides of the keyboard is as equal as possible.

Starting from the application choosing a 13-key code and biased toward right-handed dominance to illustrate its principle, the keypad design as in fig. 25 can be built. When this one-handed keyboard is used as a two-handed keyboard, the resulting design is shown in FIG. 26. On this keyboard, the right hand can type about 84 percent of letters, while the left hand can type about 16 percent of letters. This asymmetry is desirable in most cases where the key button actuation will be performed in a nearly identical manner, whether using a one-handed or two-handed keyboard.

On the contrary, if typing is performed with a two-handed keyboard and only one-handed keyboard is used occasionally, it can be said that the weight of relying on both hands is as equal as possible when operating the two-handed keyboard. This goal can be achieved through an alternative design of a two-handed keyboard shown in fig. 27. It is worth mentioning that, regardless of whether the right or left hand is used to type on this one-handed keyboard, 50 percent of the typing activity on the two-handed keyboard is substantially the same as the typing activity on the one-handed keyboard.

Also worth mentioning is that there is 2 for the 13-key ambiguous code¹³There are different ways to pair the corresponding left and right hand buttons of a two-handed keyboard. The number is small enough to allow each pair of corresponding specific gravities of the hands to be evaluated, and the appropriate setting to choose from will depend on whether the specific gravity is most symmetrical or most asymmetrical or some intermediate value is desired.

Reference will now be made to fig. 28, a detailed view of a single-handed keyboard, which shows how these objects are achieved by positioning on the keyboard a plurality of key buttons 300, a thumb-operated typing device 301, a mouse 302, a palm grip 303 and a display 304. The keyboard may further be provided with a communication device to facilitate symbol selection procedures transmitted between the calculator and the keyboard. This may be a wired or wireless communication device, such as an infrared communication device. The keyboard is securely movable on a support means, such as a table, so that the keyboard can be operated by sliding on a device pressed by the palm of the hand. It would be highly desirable to provide a palm grip on such a keyboard to effectively move the keyboard in response to known pressure from the palm of the hand. Under the condition of a mode that the keyboard can be effectively moved in any direction by only slightly pressing, the device for operating the keyboard by palm force is an ideal device. This pattern may be, for example, a groove that can be securely placed on the palm of the keyboard. By moving the keyboard in this manner, the fingers operating the keyboard can be used to effectively operate the key buttons, although the same is true when the keyboard is moving. It follows that the keyboard can be used, for example, in the case of playing computer games, i.e. when it is desired to type in a sequence of actions and symbols simultaneously.

It is noted here that the appearance of the device is very different from that of a mouse when it is performing the mouse function. The mode factor is determined by the hand structure of the touch typing in a comfortable position. Thus the device must be much larger than a standard mouse and the means to move the device will be substantially different.

To communicate keyboard activity 305 to the calculator, the keyboard is fitted with a motion sensing device, such as a trackball, as is well known to skilled users. The keyboard 305 is ideally further able to mount a bias to the device, such as a spring, to lift the keyboard from the support device when the hand pressure on the keyboard is reduced, and so is easier to move. Conversely, when substantially full hand pressure is applied to the keyboard, the keyboard remains relatively stably mounted on the support device while typing activity is assisted.

Thus, the two-handed keyboard can be used for typing without being influenced by the movement of the mouse for a long time, and the one-handed keyboard can be used for typing quickly and variably by the movement of the mouse.

It is worth mentioning that visual representation of the keyboard when using a single device that supports more than one ambiguous code or more than one mode, it is practical for the user to have a representative of the current associations of buttons and symbols appearing on the display at any given time. Such a representative would be advantageous for any typing device, especially for touch-typable devices, because some or all of the keys would be visually inaccessible (particularly for keys next to the finger of an operating finger) during use of such a device. Thus, any display combined with a keybutton would be limited in functionality for a touch typist. The most practical visual representation is in the case when the physical design of the keyboard is implemented on a visual display. Such a device 304 is shown in fig. 28 and may be incorporated into many of the applications already described herein.

Reducing typing injuries many keyboard users suffer from typing injuries (repetitive stress syndrome). Several keyboards have been designed to reduce the pressure on the repetitive motions involved in typing. The most effective method for reducing typing damage has long been recognized as the regular rest of the typewriter during typing. This is not practical, however, because typists are often time-critical in completing their typing tasks. The one-handed keyboard just described provides a solution for this. Although this current one-handed keyboard is operated with the right hand, the same design methodology is clearly applicable to left-handed operated one-handed keyboards. These keyboards may each encode all of the same symbols. A therapeutic typing device incorporating a right-handed keyboard and a left-handed keyboard, such as the left/right-handed mouse/keyboard described above, can be operated either with the left or right hand. With such a pair of keyboards, a user who wishes to reduce repetitive stress injuries can use one of the keyboards to type for a period of time, such as 15 minutes, and then switch to using the other keyboard for the next period of time, so that the user can have a rest period for each hand and the amount of typing will not be reduced. The typing device can be provided with a locking device to lock one of the keyboards so as to realize alternative use. It is worth mentioning that after the treatment is completed, the user can return to the two-handed keyboard mode without having to relearn the required skills.

Foldable Personal Digital Assistants (PDAs) it is convenient and ergonomic to place the text screen of an ambiguous typing device in a keyboard location operated by fingers of one or both hands and the thumb of the user in a position to operate another typing device, i.e., above the thumb or between the thumb and the finger operated keyboard location, in a smart card application as described above. This application is now a broad area of consideration for a typing device that uses the same concept in combination with a folding concept to design a two-fold information appliance that will ergonomically perform different functions when unfolded, folded once, and folded twice.

This two fold design is a surprising result of ambiguous codes. It is noted that the typing device constructed in accordance with the method taught in this invention enables the keyboard to be simultaneously (1) effectively encoding natural language, (2) using substantially full-size keys, and (3) sized for pocket or compact carrying. This application is based on the use of substantially identical initial components to build a palm-size computer device, i.e. a keyboard size designed with ambiguous codes, said components being adjustable in several different situations depending on the needs of the user at the time. These primary components may be connected to each other in a folded and/or removable fashion in each of the different states. Thus, the calculator device can play the roles of a notebook calculator, a Personal Digital Assistant (PDA), a telephone, a game machine and the like.

Referring first to fig. 29, we detail a two-fold calculator comprised of four substantially identical parts, each for performing a designated function and connected to each other in a folded and/or removable fashion. Fig. 29 illustrates such a device in a folded mode. It can be seen that the first side of one of the components 900 is used as a first visual display, the first side of component 901 is used as a first keyboard, and the first side of component 902 is used as a second. Ideally, the keyboard design on the first and second keyboards would be a 13 letter button keyboard, although many other options are possible. The first side of the last component 903 is used as a pair of mode-switching thumb switches for operation in conjunction with the first and second keypads herein. The first keyboard here is designed to be operated with the right hand. A similar mode exists that can be typed using the left hand, as would be apparent to a skilled user, and can be achieved through simple rearrangement and reinstallation of the four components. Indeed, a two-handed keyboard can be achieved by rearranging the four components, as shown in FIG. 33.

Shown in fig. 30 is the bottom of an unfolded two-fold calculator. Component 904 is a telephone keypad 905 and corresponding second visual display 906. Component 907 is a third visual display and component 908 is a third keyboard. The members 904, 905, 906, 907 are the second faces of the members 900, 901, 902, 903, respectively.

By folding the calculator according to the folding line 908 shown in fig. 29 and 30, we can obtain the pattern shown in fig. 31. In this mode the third keyboard is used for typing and the third visual display is used for displaying the corresponding image. The keyboard is here designed as a 12-key keyboard, but many other options are possible. This mode can also be used in situations where the user cannot or does not want to fully deploy the calculator due to time and space constraints. It may also be used to support a different functionality than a fully deployed calculator, such as a gaming function.

Finally, the computer can fold according to the fold lines 909 shown in FIG. 31 to form the two-fold mode shown in FIG. 32. This is a particularly suitable mode for portable calculators; in this mode the device is pocket sized. In addition, the functionality of the telephone is also included in this one two-fold mode. This will be the most common mode of use of the device for many users. It is noted here that the telephone set is described as using the ambiguous code of the previous application, but many other options are possible.

Declaring again: without being attributed to the ambiguous codes, it is not possible to design a portable communication and computing device that can be converted between a telephone, a personal digital assistant and a notebook computer.

Skilled users will find that these different modes and the use of the device are further expanded if each primary component, including the keyboard, is touched by the screen. However, the tactile feedback of the standard keyboard and depressible buttons is lost. Many other alternatives are possible in keeping with the principles of this invention.

Typing device software including a touch screen application the present invention enables both software and hardware to be applied simultaneously. In particular, the method of the present invention may be used to design typing mechanisms for devices including an on-screen touch, such as the PDA family under the PALM PILOT trademark manufactured by 3Com corporation and other branded products. For the purpose of this description, we will focus on the PALM PILOT family of products, including PALM-type calculators that are capable of or have implemented one of the applications, but the approach described herein is applicable to any typing device that includes an on-screen touch.

Referring to fig. 34, we note that the PALM PILOT family of devices is generally made up of the following components: a touch screen 1000, a touch sensing area 1001 for typing characters through handwriting recognition software. The touch sensitive area referred to herein may be a sub-area of the touch screen or may be applied separately.

One of the main and surprising features of this application is that when using a touch-type keyboard on a device formed by a touch-sensitive screen, we have devised a new user interface for information devices, whereby the keyboard does not have to compete with applications on the limited screen space. The same touch screen can also be used on both the application and the keyboard.

It is an important discovery that if the keyboard is of the low touch type, it is not necessary for the user to actually display the keyboard. The user's fingers "know" the location of the key buttons without visual indication. Thus, the keyboard can be used to enter data with each application displayed on the touch screen. Furthermore, if the keyboard is touch-typable, it can be used to type high quality text, even when there is no screen space available to provide query feedback to the user.

Referring to fig. 34 and 35, some of the unique functions of the PALM PILOT family of devices that have been described in this application are:

□ the touch screen 1000 can easily replace the functions of other keyboard designs.

□ the touch screen 1000 can display images with different brightness and different colors.

□ typing font area 1001 is located at a distance from the touch screen or in a non-central area of the touch screen.

□ the PDA is used to execute various programs, such as sequence programs or communication rate programs, which occupy the same touch screen space as the keyboard.

Touch screens can easily replace other keyboard design functions in this application to implement a designated key device to represent a number of different symbols or groups of symbols, depending on the keyboard mode at any given time. When the touch screen is used on a keyboard, each of the key input devices is connected to a designated area of a touch screen. The touch screen can be used as both a visual display and the dual function of multiple mechanical key devices for assigning different functions and different labels to each key device depending on the mode. However, it is worth mentioning that the same effect can be achieved with the conventional depressible mechanical key of the structure in which each mechanical key is mounted with its own display device. Thus, the mode switching method indicated herein with reference to a device formed by a touch screen can be applied to devices formed by mechanical buttons, such as many of the other devices described herein.

Mode selection one strategy for increasing the number of symbols in a keyboard design that can be coded given a fixed number of keys is to increase the keyboard with the number of mode switch keys. Pressing one mode-changing key changes the symbol encoded by a plurality of other keys. One standard example is the shift on a typical typewriter keyboard through the changing of alphabetic keys from lower case to upper case symbols. Capital letters can in principle be coded in a different set of keys than keys used for lower case letters, and it would be a desirable choice if the frequency of occurrence of capital letters in normal communication were the same as for lower case letters. Also, in principle, the lower case letters that can be reached in the different modes do not mean that an upper case letter must be connected to the same key button for the corresponding lower case letter. The same key button designation is also selected in practical use because it has the characteristics of being very traditional, easy to understand, having statistical relationship between upper and lower case letters, and the like.

It follows that the following three principles dominate the assignment of symbols to modes and keybuttons in the modes: the degree of association with statistical relationships, the degree of association with traditional relationships, and the conceptual relationships between symbols. The problems associated with modular design are exacerbated in the use of ambiguous codes in the design of typing devices because several keys must already have the capability to code more than one alphabetic symbol on each key, and the number of keys sufficient to code a symbol is usually very limited. However, in the case where the non-alphabetic symbols are closely related, such a method that has been applied to the manufacture of ambiguous codes for alphabetic symbols can be similarly applied to non-alphabetic symbols such as punctuation marks.

The symbols that need to be encoded by the keyboard are divided into several sub-groups corresponding to the modes. Depending on how many operations are required by the user to achieve the respective modes and/or the frequency of use of the symbols in each mode, the modes are at least partially arranged. Thus, we can refer to these modes as first, second, third, etc., with the increase in the amount of operations required to achieve the respective modes and/or the decrease in the frequency of symbols used in the modes.

It may be desirable to place the alphabetic characters in a single or a plurality of first patterns. Subtle design issues must be related to the method of assigning non-alphabetic symbols to patterns and the spatial design arrangement of each pattern.

The first statistical criterion to be considered is the frequency of use of non-alphabetic symbols. Some non-alphabetic symbols, such as punctuation marks and numbers, are important for communications that may occur with equal or greater usage than alphabetic symbols. These punctuation symbols are candidates for inclusion in any active keyboard design in the first or second class of symbols. This statistical criterion to be considered next is the connectivity resulting from the non-alphabetic symbol coming into contact with other non-alphabetic symbols. There is a conventional and conceptual relationship between some non-alphabetic symbols and other non-alphabetic symbols, for example, the (left bracket) symbol is connected to the (right bracket) symbol because it makes sense that the two symbols are together. The symbol is a sum, the symbol is concatenated because the two symbols have similar meanings, end-of-word or end-of-sentence. These are examples of global generality that are quite familiar to most language users, including in these cases many languages such as english. Other more native relationships may also be considered in the design of a special purpose keyboard, for example, in: and/or the relationship between: the web page address (URL or web address) commonly used in the Internet.

Non-alphabetic symbols may also have statistical, traditional, and conceptual relationships with alphabetic symbols. For some symbols, it is possible to analyze their statistical relationship to each other through a reference corpus. For other symbols, analysis of statistical relationships between these symbols requires user research reports or specialized software, since they almost never appear in text. Examples of such are backspace, page up, and other symbols used for editorial viewing or otherwise processing text.

In the case where the reference data is derived in this way, the next symbol-to-pattern assignment step will arrange the symbols in a way that best matches statistical, traditional, and conceptual relationships. A further limitation that may also be taken into account is the memory potential of this arrangement. Ideally, all symbols are "meaningful" in all modes, i.e., the symbols are simple, familiar, and preferably visually well-formed. Even a severely trained touch typist may fall back to finding infrequently used symbols on the keyboard using a visual scanning mode. In this case, the memory potential may become a major consideration in arrangements dedicated to less used symbols. It is worth mentioning that the memory potential can be quantified through experimental rules established for the purpose of secret writing tasks familiar to psychologists.

To describe this method, a standard design of alphabetic symbols [ a-z ], numbers and 32 non-alphabetic symbols is necessary

~`！@#$％^&*()_-+＝{[]|\：；”<，>.？/

This is found on a designated one of the keypads. Such an arrangement is made up of three mode switch buttons, three modes each containing 16 symbol buttons, as shown in FIGS. 36A-C. This configuration is designed for the PALM P ILOT family of devices. This design has not been proven to be optimal by psychological testing.

The first mode, (FIG. 36A), here, includes an ambiguous code of a letter, a space/backspace button, a base punctuation and a button that can shift the mode forward or backward.

Here, in the second mode, (FIG. 36B), several additional punctuation marks are included, the arrangement of these marks being 1) a shift button attached to a symbol with an associated meaning, e.g., left and right brackets, or, in the absence of an associated meaning, to the shape of the symbol, which will help to memorize the position of the symbol. A convention is applied here, namely hard symbols, more water-caltrop symbols, on the left, and soft symbols, more curved symbols, on the right. Since all or most of the key buttons have exactly two symbols, the ergonomic disambiguation apparatus as described above can operate in each mode.

The selective transparency of the keyboard currently used in a PALM PILOT series touch screen device is such that the keyboard occupies a portion of the screen, while the remaining screen area is dedicated to an application program that receives keystrokes from the keyboard, such as an address book application program. With the extremely limited display area of such devices, the coexistence of keyboards and applications results in the need for both keyboards and applications to be very compact. The keyboard used in this device is not suitable for touch typing and is not touch-typable due to its size. However, the keyboard manufactured by the method of the present invention is suitable for use as a touch key device, even though the keyboard is located in a limited position on a touch screen of a personal digital assistant. An important discovery here is that in the case of a touch-key keyboard, the keyboard would not need to be displayed to the user. The user's finger "knows" the location of the key buttons even though it is not actually visible. Thus, the keyboard may be transparent, occupying the entire touch screen, while the application may be opaque, occupying the entire touch screen at the same time. With such a display method, the user can type directly on the application program. In fig. 35, there is a keyboard and an application 1002 displayed in this way 1003, where a drawing program is shown performing drawing functions. It is not possible to draw a transparent keyboard, which is marked in grey and the application in black. Indeed, it is feasible to let the user choose the keyboard transparency, e.g. depending on her or his touch typing skills.

It is noted that the different modes may include symbols of different familiarity to the user. Taking these differences into account, the transparency of the keyboard can also be adjusted according to the mode function, and the transparency is reduced as the unfamiliarity of the symbols in the mode increases. It should be noted that the same effect of distinguishing between keyboard and application can be achieved by adjusting other visual factors, such as adjusting the color of the image in addition to adjusting the transparency.

An important discovery of the integrated coordination/ambiguous keyboard from this aspect of the invention is based on the fact that the coordination style requires only two keys for substantially simultaneous operation, such as a coordination style that encodes capital letters with a Qwerty keyboard, will be available for immediate learning and for use by a larger group of users. However, conventional approaches have proven that more complex coordination patterns than this would not be widely accepted.

It is worth noting that there are two main conventional approaches to manufacturing typing devices with a small number of keys: a coordination method and an ambiguous encoding method. One aspect of the present invention is to combine the two approaches in describing how to complement each other.

We can distinguish between two coordination methods 1) retaining a key or keys as a method of forming a harmonic function, the familiar shift key that keys capital letters through a coordinated combination of a shift key and an alphabetic key is an example, and we generally refer to such a key as a shift key, and 2) coordination methods formed by operating several alphabetic keys substantially simultaneously. The first of these methods is used in this application and the second of these methods is used in the next application.

The main focus of this aspect of the invention is that a substantially simultaneous operation of a set of typing tools by a user can be immediately combined into a single command. Thus, a group of key actions is not easier or harder to manipulate than a single key action, but a group of keys essentially covers more information than a single key, i.e., can be used to create an easy-to-operate and low-resolution code and a typing device built upon such a code. It follows that in order to enable simple operation of the keyboard, coordination must be performed for no more than one set of key buttons in substantially simultaneous operation. In this application one of the groups is reserved as a key for forming a coordination call, while the other group is a key corresponding to at least one decoded symbol. Fewer symbols may still require substantially simultaneous operation of more than two keys, and thus, common symbols may be associated with a single typing tool without exceeding the scope of the invention.

It may be desirable that at least one of the decoded symbols is a highly correlated symbol. Thus, if a consistent formation is incomplete, i.e., if the shift button should not be pressed but pressed or should be pressed but not pressed, the discrimination elimination software can be used to correct the error.

It will be appreciated that, depending on the application, a priority can be formed in any manner when performing coordination and discrimination coding, and the ideal combination is as follows

□ no more than two typing tools need to be substantially synchronized to encode any substantially possible symbol.

□ find the error rate and/or query rate to be optimal.

□ coordination may be accomplished via a mode shift key.

□ (ideally) the probability of using the mode shift key is minimal.

When combined coordination and ambiguous code procedures are accomplished in accordance with this principle, surprising and complementary results are achieved, as will be explained in this application by an application of the integrated coordination/ambiguous code procedure to telephones that include standard ambiguous codes. We have seen that the search error rate and query rate of standard ambiguous codes are quite poor. Thus, it is quite unusual to use a comprehensive tool to enable an application standard ambiguous code keyboard to be touch-typable. The object of this application is to produce a keyboard as described below

□ are easy to touch, are fully compatible with standard telephones that employ standard ambiguous codes,

□ is easy to handle, and can be easily operated,

□ are easy to learn and can be used for learning,

□ and with minimal key button actuation.

A standard telephone set using the standard ambiguous code standard is shown in fig. 38. In this figure it can be seen that there are several buttons 10000, here eight in number, for letter and number coding. The two buttons 10001, 10002 are only digitally encoded, while the two buttons 10003, 10004 are encoded with non-alphabetic symbols #and #, respectively. In this application, one of the keys selected from the group consisting of 10001, 10002, 10003, 10004 is used as a mode-shift key, ideally 10001 is coded with the number 1. The key selected will be referred to as a shift key for reasons that will become apparent. The keys 10001 can be operated conveniently with the left thumb when the phone is held in the left hand, while the right hand is used to operate the other keys. For an application where the right hand holds the phone and the thumb of the right hand operates the shift button, button 10004 can be used as the shift button.

10000 each has a corresponding letter, and these letters are divided into two sub-groups, which we will refer to as a transition group and a non-transition group. The designation of the letters and key sets (transition set, non-transition set) is as follows

□ the lowest error rate of the search,

□ the lowest query rate of the queries,

□, the transition group, without loss of generality, will have a letter-to-key character,

□ lowest operation probability of shift key button

In general, optimizing synchronization based on search error rate, query rate, and other ergonomic criteria means a compromise. For example, it is possible to achieve a better search error rate and query rate by eliminating the limitation of one of the groups of keys (transition group, non-transition group) that each key button has a single letter so that the number of letters in the transition group can vary from key button to key button. However, the regularity of the separation between the transition group and the non-transition group makes the keyboard easier to learn and comply with ergonomic standards, which is a priority here. Learning can be further improved by selecting individuals that are also easy to remember when the number is aggregated or by memory. However, this one choice may affect the lookup error rate as well as the query rate.

A total of 11664 different pairings of transition/non-transition groups are limited to the fact that each key button of the transition group contains only one letter. This number is small enough for its ergonomic properties to test all possible combinations.

FIG. 39 shows the results of testing all of the 11664 codes and Standard Ambiguous Codes (SACs) described above as a plot of search error rate versus query rate. It is worth mentioning that while all codes are better than the standard ambiguous code, most have only a small degree of superiority. However, this is a very large distribution, in terms of the optimal encoding CEHLNSTY, with a lookup error rate of once per 431 characters, and a query rate of once per 21 characters, which is 15 times better than the standard ambiguous code in terms of lookup errors and 10 times better than the standard ambiguous code in terms of queries. The best code is AbC dEf gHi jkL mNo pqrS Tuv wxYz, and the parts in which capital letters are used are the constituent elements of the shifted-in code set. It is worth emphasizing again that this coding is the best coding for our existing reference data. Although this encoding is still a few two from many other english language corpus data, other data may still yield different optimal encodings. The generalized type coordination/ambiguous encoding method can be applied to any ambiguous code, and the ambiguous code is not limited by the standard ambiguous code or the English letter arrangement, nor by 8 keys or the equal separation method as much as possible. If the coding has more free choice space, the quality of the generalized co/ambiguous coding method can be greatly improved in terms of search error rate and query rate, and other ergonomic criteria, such as minimization of shift key usage, can be advantageously combined with the optimization of error rate and query rate. Such an optimization step is aimed at the full compatibility of existing telephone systems. Beyond the scope of the above applications, the present invention is also intended to cover such applications.

To facilitate the learning and operability of the keyboard, it can be seen in FIG. 38 that the letters comprising the transitional code group are presented in uppercase on the corresponding keybuttons, while the non-transitional code group is presented in lowercase. Or different sizes, colors and fonts are used to indicate the difference of letters.

The operation of the device is no longer simple. If a letter of the escape code set is included in the text to be typed, the letter will be presented in a non-discriminatory manner if the escape button must be pressed simultaneously with pressing the corresponding button. Conversely, if a letter in the non-transitional coding set is to be typed, pressing the corresponding key button will cause the letter to appear in an ambiguous manner.

It is seen in accordance with the teachings of the present invention that the disambiguation function corresponding to this application may be physically incorporated into the phone body. Can be used at the transmitting end of the communication and/or at the receiving end of the communication, such as a central computer contacted by the user by telephone.

There are 4 keys on a standard telephone keypad that can be used to encode non-alphabetic information, such as mode change functions. Using the composite coding symbology described above, the number of symbols used to code non-alphabetic information on a telephone keypad may be further increased. The number of transferred letters can be increased slightly, especially if less dissimilarity is desired. For example, by linking each of the 4 shift keys and one of the letter keys, the character can be typed in a completely different manner even though the number of keys required for each letter is increased. In short, the connection of the shift key and the symbol decoding subset can have many imaginations. More specific discussion of internationalization of the teachings is provided below, and so far only the details of the english part are described.

It is worth mentioning that the shift key and the remaining non-alphabetic keys arranged as the # and 0 keys may encode at least 6 non-alphabetic keys, such as punctuation marks, mode switching marks, etc., in case of simultaneous use.

Error correction using standard ambiguous codes the keyboard for this application can still encode the intended text when the shift button is pressed without or without pressing the shift button, especially when used by a novice. Such an operation will typically result in a meaningless code in the case that the disambiguation means codes correctly for the integrated coordination/discrimination code. When this occurs, and the query is issued, the code sequence can be interpreted as if it is considered the same as the standard ambiguous code by trying other methods of disambiguation whether or not the shift key is pressed. Typically, such interpretation will locate the text that the user originally intended to type.

It is to be noted that in the apparatus shown in fig. 38, a high-relevance symbol (space bar) and a low-relevance or non-relevance symbol (elimination bar) are paired. This pairing method is essentially intended to allow the ambiguity resolution software to correct errors such as the pressing of the space key with the resolution key or the pressing of the space key with the resolution key.

Adapting touch typable query methods when querying is permitted as described above, it may be desirable to use the shift key as the scroll key. It is worth mentioning that the key button can be positioned for the function of the shift or scroll at any time as long as there is suitable software. The key functions as a scroll when the device is in the inquiry mode and as a transition in other cases.

The alternative positions of the shift key buttons are again referred to in fig. 38, knowing that this application is designed to operate with existing standard telephones. If the phone is manufactured with this application as a premise, key button 1005 should be added as a function of the transfer. These additional keys should preferably be arranged on both sides of the phone so that the keys can be activated with the thumb when the phone is held in the palm of the hand. Fig. 38 shows the above arrangement. Other keys 10006 that can be activated by the fingers of the same hand (either left or right) can also be used.

Low frequency character screening in queries typically distinguishes between a very high frequency character and a very low frequency character. For example, in the case of CEHLNSTY in English, the very high frequency character "for" is distinguished from the very low frequency character "fop". Removing these very low frequency characters from the dictionary can effectively improve the query rate with minimal impact on the dictionary in the language it represents. For example, deleting characters less than one fiftieth probable may improve the query rate for CEHLNSTY to occur every 46 characters. Such screening can be achieved, for example, by a "gap factor". The "gap factor" is then the ratio between the two characters being queried, e.g., the ratio of the highest frequency and lowest frequency characters. For example, assuming a gap factor of 500, we obtain a distribution as shown in the xxx, and the two codes specifically indicated in the xxx are the standard discriminatory code (SAC) and the CEHLNSTY code, respectively.

Internationalization is the key to internationalizing the application, which is easily solved by users familiar with the teachings of the present invention. These two points are 1) the processing of the pitch, and 2) the creation of inductive codes that can be used in multiple languages simultaneously. These two points will be briefly discussed later.

And (4) processing the cavity tone. Many languages use letters containing the key of yankee cavity. For example, "e" in French may have a notation of "e", or "e". If there is no cavity tone suppression, "efeve" (meaning "student") will diverge from "efeve" (meaning "rise"). We can proceed from the use of another shift key. When the shift key and the code key are used simultaneously, the key having the function of selecting the letter containing the cavity key can be called as the cavity key shift key. For example, when the French uses CEHLNSTY coding, the Deliver cavity transfer button can be used with the def button to code either the de or e, and the ambiguity resolution function can be used to determine which is the appropriate Deliver cavity modulation. Using the above steps, we find in a set of french character frequency data that the (search, query) rate encoded by CEHLNSTY is (38, 3) when the wife-no cavity shift key is used, and (584, 24) when the wife-no cavity shift key is used. The bottom row of keys on the telephone keypad may be used as a function of the cavity transfer. On the special key button plate, the key buttons can be additionally added to provide the function of inhibiting the transfer of the cavity. From an ergonomic point of view, it would be desirable if the cavity shift key could be operated with similar performance to a conventional shift key. For example, one direction of thumb motion may be used to encode the operation of a normal shift key, while the other direction of thumb motion may be used to encode the operation of a cavity shift key.

Since different languages usually have different data, the codes that are optimized for one language are not necessarily optimized for the other languages. For a certain language, a code that is typed with high touch key strength cannot meet the requirements of other languages.

For example, CEHLNSTY, which is optimized for english, performs poorly when used in french for coding selected bitwise for french optimization. Although CEHLNSTY is a high degree of touch in french typing, this is not the case in every language.

To achieve an economic advantage, manufacturers should produce a machine that is adaptable to a variety of languages in the region. For a key-typable device such as a mobile phone, the keys are marked with ambiguous codes designed for them. If a code is selected that can be applied to multiple languages, the key tag method can be used on all production lines regardless of the specific language area.

By applying the above-mentioned techniques, it is possible to compile ambiguous codes optimized for different languages simultaneously. In the method for multi-national language optimization, the step depending on the weight between ergonomic standards may be counted as one of the sub-steps of the step depending on the weight between multi-national language standards. Different specific gravity modes must be used in different situations. For example, in optimizing both English and German data, it is likely that the code will show more weight in English than in German.

Maximizing the lowest performance would be the most desirable method of specific gravity. The above steps are called the lowest-highest steps.

Suppose that a set of different languages 11, 12, are to be optimized, i.n, and a set of ergonomic criteria e1, e2, em. Given the ambiguous codes c1, c2 and the use of em as ergonomic criteria, c1 has a higher rating than c2 if the minimum value of em divided by the language ln is higher for c1 than for c 2. When there are many ergonomic standards that must be optimized, it is possible to find a code that is particularly good under one particular ergonomic standard and relatively bad under another particular ergonomic standard. For this example, the ergonomic criteria must be considered in terms of importance as a weight ratio with respect to each other, as described above.

In accordance with the above concepts, for example, assume that there is a set of languages that need to be optimized for lookup error rate and query rate. In this example we will use 8 standard typing functions, one arbitrary typing function, and an arbitrary typing function of the cavity transfer key button, in a non-alphabetical order. Thus, the 8 standard key functions can be activated simultaneously with any one of the arbitrary key functions as described above to achieve the application of comprehensive coordination/discrimination coding.

First, the optimized language set is considered to include French, Italian, Portuguese and Spanish, with separate reference data for each language.

The following encoding is readily obtained using guided random marching: joz m bhx akn r pw d iy l gq t ev c fu s. The (search error rate, query rate) for French, Italian, Portuguese and Spanish are (3250, 265), (11400, 3800), (4720, 505) and (6280, 400), respectively. The codes perform poorly in Dutch, English and German languages, namely (65, 4.8), (93, 10) and (360, 13).

The optimization of Dutch, English and German characters in the same time spent computing the calculator can obtain the following codes: cjk r biy l fv e mo a sz p hx g tu d qw n. And (lookup error rate, query rate) are (1220, 44), (816, 44) and (480, 47), respectively. The (search error rate, query rate) for French, Italian, Portuguese and Spanish are (253, 20), (306, 50), (525, 36) and (4236, 272), respectively. Although the above results are just good for these randomly sampled languages, they are still very different from the results specifically optimized for language ambiguities. The above results indicate that the greater the difference between each language, the less common the coding performance.

In a real world scenario, whether to include certain languages in a multi-lingual optimized plan is a business rather than a conceptual decision. The emphasis of this invention is on the ability to provide high touch typing capability even with minimal expression in the selected language. In the previously examined examples, the optimization for French, Italian, Portuguese and Spanish produces C-level high-touch codes, but the least expression for Dutch, English and German is only A-level high-touch.

One-handed, high-touch handheld devices have been found to be useful in the above-described generalized co-ordination/ambiguous code applications in that superior keying capabilities can be used for ambiguous code symbol encoding to reduce overall system ambiguity. The above prior art applications show that the same set of keying functions can be used to coordinate the encoding of information and ambiguous code symbols. The ambiguous codes in the above example can be represented using multilevel coding: the keyed sequence of the first segment is used to select a first subset of codes, while the keyed sequence of the second segment is used to select a second subset of codes, and so on. Ideally, the second subset is one of the first subsets, and the third subset is one of the second subsets (i.e., one of the first subsets). This is the "tie-point occupation" method familiar to pedestrians. So far it is not fully understood that a) the number of consecutive subintervals in a group of symbols may be limited by the minimum subset containing more than one symbol, becoming a discriminative code. b) The discrimination of the last ambiguous code is minimized by the choice of the sub-separation property. c) The discrimination elimination can be optimized while simultaneously improving other ergonomic criteria such as when following traditional. d) The grade-up grade transfer can be realized by only pressing the key buttons matched with the lower grade.

Specific extensions to the above findings will be discussed in more detail with reference to fig. 39-47. It should be noted that this is one of an unlimited number of devices that can be manufactured in accordance with the teachings of the present invention. All keying devices that use the point-tie method to create codes and improve other ergonomic criteria simultaneously, such as optimizing the keying device while following traditional practices, are included in the above application.

The application device has a high degree of key touch available with one hand. Additional advantageous features include:

1 the usability of completely nondifferential text entry methods.

2 optimizing the usability of the ambiguous text entry method with high touch sensitivity.

3 usability of lowest key number mode in data capturing.

4 and above three modes have the highest compatibility conforming to the human engineering.

Ideally, in addition to the above ergonomic criteria, scan time is one of the criteria that needs to be optimized. The next paragraph will discuss the optimization of the scan time.

To create a multi-level ambiguous code based touch-enabled device, we will refer to FIG. 39 for an overview discussion.

In a first step 150, a set of second level decoded symbols must be picked. These decoded symbols are represented by ambiguous codes and may also include, for example, letters a through z in English.

In a next step 151, an ergonomic criterion for all multi-level codes has to be picked. The ergonomic criteria may be, for example, high degree of touch or a search error. Typically many ergonomic standards for all multi-level codes can be applied simultaneously. In a next step 152, the sorted second level decoded symbols are divided into subsets. Each subset of the second level is assigned a decoding symbol, so that the overall encoding can be optimized ergonomically. This approach is not different from any of the most differentiated coding approaches so far. However, there may be additional limitations, such as a limitation on the number of allowed code symbols, which also results in the performance of the next step of the method. In this next step 153, the second stage code symbols are collected into groups. These groups are treated as decoded symbols for the first stage encoding. In the absence of any unexpected condition, the encoded symbols of the second stage encoding become decoded symbols of the first stage encoding. Thus, the first stage encoded symbols are assigned to each group, which results in a first stage ambiguous code. Other optimizations that meet ergonomic criteria may be made in the assignment of second stage codes into groups. Generally, each of the multi-stage codes may be optimized according to different ergonomic criteria. These criteria may be the same or different from the ergonomic criteria used for all multi-stage coding optimizations. In a final step 154, the multi-stage encoding is applied to a typing device.

In this description of the multi-stage encoding method, the second stage encoding method advances the method to the first stage encoding. In operating an application multi-stage encoding device, this order would be reversed: firstly, a component of the first stage code is selected by operating the key-in device, and then one component of the second stage code is selected by further operating the key-in device. This is the essence of this point occupation method. This method can also be applied to third and higher level coding, as will be apparent to the skilled user.

In practical applications, the properties of each stage in the multi-stage encoding must be optimized simultaneously to achieve all the desired multi-stage encoding properties. This application at present may be specific to how this synchronization optimization is planned and performed. In short, we can go through fig. 40 to illustrate this architecture method in this application. In order to be able to display this program completely, we select three ergonomic criteria for all the multi-stage codes, the first stage code setting two criteria and the second stage code setting three criteria, thus constituting the multi-stage code. In this application, the three ergonomic criteria for multi-stage encoding are key-touch sensitivity, query rate, and search error rate. The first stage of coding is optimized according to the structure-searching precision and the English letter sequence arrangement, and the second stage of coding is optimized according to the separation equality, the structure precision and the substantial English letter sequence arrangement.

The first step 3100 in this application is the selection of (second stage) decoded symbols. I.e. from the letter a to the letter z. Next, steps 3101, 3102 and 3103 select the degree of easy touch, the query rate and the search error rate, respectively, as ergonomic criteria in the multi-stage encoding. Next, the structural accuracy is selected as an ergonomic criterion in step 3104. Since the device is designed for finger-typing with a hand-held device, the structural accuracy is at its highest, i.e. 4 keying-in devices and 4 corresponding first-stage code symbols are each assigned to a different finger.

The structural accuracy is selected as an ergonomic criterion in a second stage step 3105. Each code symbol in the first stage code will correspond to several second stage code symbols. If each of the 4 first-stage symbols corresponds to 4 second-stage encoded symbols, the structural accuracy of the second-stage encoding can be maximized, i.e., the structural accuracy is maximized in the case of 16 second-stage encoded symbols. If 26 second-stage decoded symbols are distributed among the second-stage decoded symbols and 1 or 2 of the second-stage decoded symbols are concatenated with each of the 16 second-stage encoded symbols, then the 16 second-stage encoded symbols may also be concatenated with the second-stage decoded symbols, thereby achieving the best degree of separation. This distribution means that there will eventually be one connection between 4 and 8 second stage decoded symbols to each of the 4 first stage encodings.

Next, in step 3106, we select English alphabetical ordering as an ergonomic criterion in the first stage encoding. Optimization to meet this criteria requires simultaneous optimization of the first and second stage codes. That is, the letters a to z need to be arranged in the order of the english alphabet to be displayed on the display corresponding to the key input device connected to each finger. Since the displays are arranged in finger order, this arrangement means that the first part of the letters in the alphabetical order must be connected to the second stage decoding symbols connected to the first stage encoding symbols connected to the typing device to which the first finger is connected. Similarly, the second group of letters, following the second group of letters, must be connected to the second-stage code symbol that is connected to the first-stage code symbol connected to the key input device to which the next finger is connected, and so on for the other two first-stage code symbols. Thus, optimizing by letter order corresponds to selecting a permutation bin for the 26 letters, i.e., in the same manner as discussed in other applications of the invention. This time, each of the 4 components in the arrangement cell must have 4 or 8 sub-parts, so that all the ergonomic criteria we list can be optimized simultaneously. This will be the same as shown in the detailed description of the best mode of application, i.e. we can find the most equally spaced codes possible when optimizing against all other ergonomic criteria considered, which is also the case for the first stage codes.

Finally, in step 3107, we select the substantially alphabetical ordering as an ergonomic criterion in the second stage ambiguous code. This means that we will arrange the letters in alphabetical order as much as possible given that all other letters have known limitations on the assignment of second stage code symbols. The deviation from strict alphabetic ordering can be measured by a number of methods, for example, by substituting a known arrangement into a strict alphabetic ordering by the number of pairs required.

Referring now to fig. 41-47, a touch-typable device for single-handed typing, encoding at least the letters a-z, and incorporating an encoding constructed in accordance with the method described above, will now be described. To make the device touch-typable according to the method of application, it is necessary to fix the separation of the symbols into subgroups, subgroups within subgroups etc., that is to say, for example, not to change depending on which symbol was previously typed. This fixed separation is only suitable for the required degree of easy-touch keying, however the principles of the invention can be applied in a wide range. For example, the number of key button actions can be greatly reduced using a character completion mechanism. The behavior of the character completion mechanism is complex and not easily predictable, and devices with character completion functions do not become touch-sensitive. However, the same optimization procedure for easy-touch encoding can achieve an efficient character completion mechanism, with less ambiguity and improved character completion. It is known that adding a character completion mechanism to a touch-typable device does not allow the device to fall outside the scope of the present invention.

We now place further limitations on this application, namely that the symbol-typable portion of the device must be able to be held in one hand, and must be typable only in the hand that holds the device. To limit the need for digital action, most symbols can be typed through a series of operations on the following 5-key devices: 4 keys operable by the fingers of the hand holding device 2100 and 2103 and 1 key operable by the thumb of the hand holding device 2104. This device, shown in fig. 41, is designed to be held by the left hand; it is obvious that a two-handed device can be designed that is either symmetrically held by the right hand or operated by one of the hands.

Ideally, associated with each of the entry devices 2100-2103 is a visual display 2106-2109 showing the subset components currently associated with known entry devices. The keying-in device is operated to select the corresponding subset. Typing means 2104 may be used to further improve subgroup selection and/or to select subgroups of other symbols. For example, the single symbol "space" can be connected to typing device 2104; this or other symbols associated with the typing means 2104 may preferably be displayed on the display 2110. The letters [ a-z ] may be distributed among the 4 key-in devices 2100-2103. Preference is given to such a distribution of letters of the typing device in order to minimize ambiguity (search error rate and/or query rate) while complying with the conventional alphabetical ordering. Such adherence can help a novice user to find a desired letter by simply scanning candidate letters.

Fig. 42 illustrates an arrangement of letters a-z, in which the letters a-f are connected to a first key input device 2100, [ g-l ] to a second key input device 2101, [ m-r ] to a third key input device 2102, and finally, [ s-z ] to a fourth key input device 2103. These concatenations form the first stage subgroup in the first stage encoding. In short, ideally, one would choose to connect 4-8 letters to each of these 4-keyed devices: the letter subsets associated with each typing device may be further subdivided into 4 subsets, each of the 4 subsets not containing more than two letters. The limited functionality will be quickly clarified and it will be apparent to a skilled user how to extend the application principle to languages containing different numbers of symbols and different numbers of typing means.

An example set of second stage subgroups that separate the first stage subgroups as shown in fig. 42 is illustrated in fig. 43. FIG. 43 is a table with four columns and four columns. The columns are indicated by the keying devices enabled in the first step and the columns are indicated by the symbols associated with each keying device in the second step. Thus, for example, if typing device 2100 is first enabled, then in a second step ac the symbol is connected to typing device 2100, be the symbol is connected to typing device 2101, and so on. In the case where the limits of the subgroup size and the query rate are as described above, this arrangement is chosen to minimize the search error rate and the query rate. Using our reference data the search error rate and query rate for this code is (1100, 69). It is important to note that in this example, the letters in the first stage subgroup are arranged alphabetically, but the letters in the second stage subgroup can only be partially arranged alphabetically. For this example, we have decided to relax the alphabetic constraints in the second stage, thus improving the search error rate and query rate, and producing a code that is as touch-typable as possible. This shows that the alphabetic ordering may or may not be optimized, as with any other ergonomic criteria, and that the optimized attribute may be different at each stage in a multi-stage ambiguous code. Once again, the advantage of english alphabetical ordering is that the scan time is reduced, especially for novice users. Since the number of symbols displayed in the second stage is small, the scanning time is originally short, and the mechanism to be discussed now can be further shortened.

To type a desired letter, the user first activates one of the key devices 2100 and 2103 corresponding to the first subset of the desired letter. Next, the user selects one of the second stage subsets again by enabling one of the typing devices 2100 and 2103 corresponding to the first subset including the desired letter. Fig. 44 illustrates an example of the operation of one such device, where the user types the letter e. Referring to FIG. 42, it can be seen that the letter e is connected to the typing device 2100 through the second stage mode. When the user activates this one key input device, the display condition will be as shown in fig. 44. The letter e is now connected to the key input device 2101. When the key-in device is operated, the letter e is output. The same sequence of operation of the keying means can be used for the selection of the letter b, so that the code is ambiguous. As in other applications, the letter b or e will depend on its context through a disambiguation mechanism.

Characters are typed through sequentially selected letters according to the method, and termination of the character is accomplished by activating a typing device 2104 connected to the thumb and hands. It is noted that this keying-in device forms the basis for a one/two handed application when a two-key method is used to code each letter. More specifically, if one hand is used to mark a first key actuation for each letter and the other hand is used to mark a second key actuation for each letter, then the messages of the first and second key actuations can be entered simultaneously. Many practical applications may be based on this. For example, the finger mechanism of [4] can be used as the practical basis for a one/two-handed application. [4] The proposed code is based on the ability of the motion sensor to sense several different positions of each finger to distinguish the disambiguating code for each letter. This would require a more accurate inductor. However, if a two-handed version of this application is used, a simpler sensor may be used. These sensors only need to record binary (up/down) information for each finger. The complexity of the software and hardware can be reduced at the same time. Additionally, devices constructed in accordance with the principles of the present invention may be easier for a user to learn and operate.

The visual cache scan time is the time to visually find a desired letter from a group of letters. Typists who do not press the regular typing method can visually scan the keyboard to find the next letter and then press the corresponding key button. The scanning time depends on several factors, including the user's familiarity with the appearance of the arrangement of the keyboard. A typist who is not in regular typewriting might know roughly where the desired letter is and only confirm it by visual scanning or find the correct position. Since the scan time improvement is achieved by the familiarity of the average user with the english alphabetic ordering, this application selects the english alphabetic ordering as the first stage encoding. In yet another alphabetical arrangement, certain letters are selected from a group of letters on a given key button displayed in a selected area for clear identification on a visual display associated with the key button. The letters are the most likely letters to be selected in any case, and placing them in a clearly distinguishable position makes them easier to find. The principle is similar to that used in some computer processors in which recently used data is stored in registers to facilitate quick retrieval when needed next time, based on the assumption that recently used data will be more likely to be reused. Here, where the language-dependent data is known, the letters are not placed in the cache because they were recently used, but rather because of the probability that they will need to be used the next time. However, visual caching of this proprietary word is still applicable.

We will now describe one application of visual caching in terms of this application's context. It is noted that the present invention allows a greater range of adjustments without having to make adjustments to adjust its primary characteristics, such as cache size and location, how caches are placed, how caches are marked, and the like.

From our standard data analysis, we find that the letters a-f, which are most likely to be the first letter of a character, are connected to the typing device 2100 through the first stage encoding. Similarly, among the letters [ g-l ] connected to the key input device 2101, "i" is the first letter most likely to be a character, "o" is the most likely one among the letters [ m-r ] connected to the key input device 2102, and "t" is the most likely one among the letters [ s-z ] connected to the key input device 2103.

By arranging the letters a, i, o, t at a distinct position in the screen, for example, at the upper left corner of the screen where each key input device is connected. This will make these letters the first contact letters of a standard visual scan from left to right, top to bottom, on one of the screens associated with each. Ideally, other choices of single letters than this alphabetical arrangement of english letters are possible, such as the alphabetical arrangement reserved for other letters in the subgroup. The distinction between letters and other letters in the cache can be further marked by selecting a font color, size, style, etc. for the cache letter that is different from the other letters.

Referring now to fig. 45 and 46, it will be seen that this finding can be used to reduce scan time. FIG. 45 illustrates how the "think" character is typed without using a visual cache, while FIG. 46 illustrates the same character with a visual cache. Thus, in fig. 45, the letter "t" is typed by first enabling the letter "t" corresponding to the first typing device 2103. The screen display is as shown in the second column of the figure before the first key entry device is enabled. When 2103 is enabled, the screen display becomes the same as that shown in the third column. When the key device 2101 is activated, the letter "t" is the output result. The screen display will change similarly when the other letters of the "think" character are typed.

In FIG. 46, the identification of whether a letter is stored in the visual cache is indicated by capital letters for the letter stored in the visual cache, and by lower case letters for the letters not stored in the visual cache. From our reference data we find that none of the 42 percent characters have a, i, o or t first letter. Thus, 42 percent of the time a user begins typing a character will find the desired letter in the cache in time.

When a character is typed, the most likely next letter will have a change in context resulting from typing the character. It follows that the letters selected for storage in the visual cache will change depending on the character being typed, and also depending on which character is typed.

In the example of the "think" character, the letter "t" can be found in the visual cache both before and after the first typing device is enabled and before the second typing device is enabled, as shown in the first four columns of FIG. 46, each corresponding to one typing device display. Once the letter "t" is selected, the letters in the visual cache become a, h, o, w, as shown in the second set of four rows in FIG. 46. After selecting the first typing device (typing device 2101) as the starting character "h", there are only two possibilities for the letters to be formed into part of the character, based on the reference data, namely the letter "h" and the letter "i", which are both in visual cache in the second display. Proceeding in the same way with the letters i, n, k, we find that the desired letter is always in the visual cache of this character. Indeed, after the first typing means is activated to type the letter "i", in the case where the user is actually typing a character in the library, there are only two letters that may be needed. Then, in the example, in the case where the second typing device is in the same virtual position, the letter "i" can be output immediately after the first typing device is enabled.

As has been described, the complete disambiguation process and the entry of other symbols generally provides a completely non-discriminative method for entering symbols on an ambiguous code typing device. In this current application, a simple way to provide non-ambiguous entry is by providing an additional non-ambiguous entry device 2105 as shown in figure 42 in a position where the thumb can be easily manipulated, i.e. in a preferred position. However, other locations may be selected.

In this application we have chosen to limit the size of the second stage subgroup to at most 2 symbols. Thus, the disambiguation mechanism will always be able to correctly select the desired letter, or will always select another letter that is incorrect but paired with it. In the case where the corresponding key-in device is selected by the user, any ambiguity resolution soft key can produce a signal indicating the two symbols it will select. This signal can be used to provide feedback to the user, for example, by highlighting (highlighting) the selected letter. If the selected letter is not the desired letter, the user may choose to activate the complete divergence elimination apparatus 2105 of FIG. 42 to force the selection of another, non-highlighted, symbol.

An example of using the non-ambiguous text entry device is shown in FIG. 48. In FIGS. 46 and 47, how the character "think" is typed is illustrated. Here, in the case where the non-ambiguous key input device 2105 is activated after the first and second key input devices for inputting the letter of the character "think" are activated, the fourth column shows the letter to be output. For example, assuming that the character "think" is typed by activating the typing device 2103 and then activating the typing device 2101, the letter "u" is selected by further activating the typing device 2105. The letter "u" will then become the first letter of the character. All possible letters can be typed in this way with the ambiguity eliminated. When this second stage subset contains only one letter, this letter will be available for ambiguous typing without activating the typing device 2105, i.e., the typing device 2105 is not suitable for this purpose. The child and mother d, f, h, l and p are always typed unambiguously according to the assigned code.

Degree of easy touch: the measurement of the degree of tactile sensitivity and thresholds is a new and inventive concept to illustrate the exact specification of the device. The breadth of this concept has been illustrated by a number of applications, namely by challenging its limits to show its scope of application.

To further increase certainty in the description of the degree of easy-touch, this section sets forth an additional numerical attribute of the degree of easy-touch that will make the degree of easy-touch of any ambiguous code measurable, thereby determining whether the code falls within the scope of the invention.

Linguistic data as we have mentioned earlier, a linguistic corpus is used to represent a research topic for a linguist. With numerical certainty, we define a representative language corpus as a total set of at least ten million characters randomly selected from a general tabloid of the target language.

The number of keys we need to define the following four key numbers: a number of virtual keys, a number of coordinate keys, a number of active keys, and a number of hybrid active keys. Number of virtual key buttons: the keys used to code the symbols are pressed several times. A most basic qwerty keyboard has 26 keys each labeled with a letter, a shift key, and a space key, such that the number of key buttons is 28. Number of coordinate keys: the number of specific combinations of key buttons for coding symbols. In the most basic qwerty keyboard, the shift key can be combined with any alphabetic key to form a capital letter, i.e., the keyboard has the coordinate key number of 28+26+ -1 ═ 53, because the shift key alone cannot be used to encode any symbol. Without exception, a keyboard identical to the qwerty keyboard may be composed of 53 physical keys, each encoding a single symbol, whether an upper case or a lower case. Indeed, some early typewriters were of such construction.

Number of effective key buttons: when a set of symbols, which can be represented by p in an ambiguous code, in a set of linguistic data and in a plurality of physical keys, exists, an optimal ambiguous code with the lowest possible search error rate and query rate is present, which is the only constraint on the code. We will easily represent the two data as pl and pq. Any ambiguous code on any number of physical buttons will have a valid button number p when its lookup error rate and query rate are equal to pl and pq. A keyboard with a substantial number of keys less than p may not support an ambiguous code with an effective number of keys equal to or greater than p. Where the effective number of keys for an ambiguous code on a physical key is less than p, it is entirely possible, and often the case, for an ambiguous code on a physical key p. Number of mixed effective keys: this is a discovery of the nature of experiments in which the search error rate and query rate of the substantially optimal ambiguous code are substantially linked by the power law, for example, by the experimental results of FIG. 11. The description herein is based on English, and the logarithm of the substantially optimal query rate is directly proportional to the logarithm of the substantially optimal search error rate.

We can find a single number that defines the lookup error rate and the lookup rate of the code in relation to each other: the location of this code (lookup error rate, query rate) is indicated on the log-number best link.

For example, standard ambiguous codes are considered. The location of this code (search error rate, query rate) is (29, 2.2). By crossing the positions linearly on the best connecting line in FIG. 11, we find the value of 5.96, which is the mixed effective key number of the standard ambiguous code. Although the standard ambiguous code is defined over 8 physical keys (the number of coordinating keys is also 8 since coordination is not required here), it is equivalent in ambiguity to a substantially optimal code of 5.96 physical (or coordinating) keys. Indeed, a few physical keys are not practically feasible, but these results show that it is possible to find a substantially optimal 6-key code that has a search error rate and query rate that is better than the standard ambiguous code.

These considerations enable us to define an accurate, although disciplined, numerical threshold for substantial optimality of the hybrid lookup error rate and query rate: if the combined effective key number is within 0.01 of the coordinate key number, a code can be said to be substantially optimal after reference to these data, in the case when no other ergonomic limitations are imposed on the system. We can also define an accurate, although also decisionless, easy-touch threshold: an english ambiguous code is defined as the most highly accessible when the combined effective key number is at least 10. We can extend this definition to other languages by limiting the need for a code to be touch-typable, and the search error rate and query rate to be greater than or equal to an english touch-typable code. However, the combined effective key count of the standard ambiguous code is less than 10, and thus is not touch-typable by consideration in this section.

By measuring the number of mixed valid keys, any ambiguous code is filtered by whether it has the property of easy key touch. For example, the integrated shape co-ordination/discriminant encoding discussed above: ab c df e gi h jk l mo n pqr s uv t wxz y has 9 essential buttons: the standard telephone has 8 keys on the keyboard and one additional transfer key. The number of the coordination key buttons is 16; i.e., equivalent to the ambiguous code of 16 independent keys minus one shift key. Without applying a spacing factor, the (lookup, query) rate would be (431, 21) and the corresponding would be a mixed-valid key number of 12.8. In the case of the spacing factor of 500, the data is raised to (440, 46), which corresponds to a mixed effective key count of 13.75. Such a code would be touch-sensitive whether the spacing factor is considered or not. It is noted that the number of mixed-active keys is less than the number of coordinating keys, but the code is substantially optimized because of the additional constraint of alphabetic ordering. Similarly, an application of one-handed integrated configuration coordination/ambiguity codes would have a (lookup, query) rate of (1100, 69) and the resulting code would have a mixed effective key number of 15, although the actual key number would be four and the coordination key number would be 16. This is a touch-typable code, and the difference between the co-ordinated key button number and the combined effective key button number is caused by additional ergonomic limitations on the alphabetic ordering of the chinese and english letters in the first stage of the code. Taking this additional constraint into account, the encoding is substantially optimized.

In contrast, the 14-prime key code of Fujitsu, pn gt cr zk wj a ehi so od xf ym vl qb, has a (lookup, query) rate of (105, 4) and a mixed effective key number of 8.47. This code is neither substantially optimized nor touch-typable, although its number of substantial keys is greater than 10.

Although an example of the use of the present invention has been described herein with reference to the accompanying drawings, it is to be understood that the invention is not limited thereto, and that many other variations and modifications may be effected therein by a skilled user without departing from the scope of the invention.

Reference to the literature

[1]http：/www，fujitsu.co.jp/hypertext/news/1996/Jun/ms-txt.html

http：/www，fujitsu.co.jp/hypertext/news/1998/May/27-e.html

http：/www，fujitsu.co.jp/hypertext/news/1996/Jul/text.htm

[2] US5,818,437, reduces the keyboard bias calculator. US58184371998 day 10 and 6

[3] Reduced keyboard disambiguation system, PCT/US 98/01307; WO98/33111

[4] "main part conjoint fortune indicates": wireless wearable keyboards, CHI 97 electronic publications: paper, author: fukumato, Masaaki and TONOMURA, YoshinobuNTT human interface laboratory, l-1 Hikari-no-oka, yokokuka-shiKanagawa-ken, 239 JAPAN (JAPAN) web reference data:

http：//www.acm.org/turing/sigs/sigchi/chi97/proceedings/paper/fkm.htm#U21。

Claims (77)

1. An apparatus for touch-key entry of text on a handheld device, comprising:

a plurality of keys;

a plurality of characters;

an ambiguous code, said character associated with said plurality of keys according to said ambiguous code; and

ambiguous code elimination;

the method comprises the following steps: inputting one of the plurality of characters according to the erasure coding when one of the keys is activated,

the ambiguous code is characterized by being touch-typable.

2. The apparatus of claim 1, wherein said ambiguous encoding is optimized for lookup errors.

3. The apparatus of claim 1, wherein said ambiguous encoding is optimized for query errors.

4. The apparatus of claim 1, wherein said ambiguous codes are optimized for lookup errors and query errors.

5. The apparatus of claim 1, wherein said ambiguous codes are optimized for alphabetical order.

6. The apparatus of claim 1 wherein said ambiguous codes are optimized for retention of a standard keyboard layout.

7. The device of claim 1, wherein said ambiguous codes are optimized for a number of keys.

8. The apparatus of claim 1, wherein said ambiguous codes are optimized for a plurality of constraints.

9. The apparatus of claim 8, wherein the plurality of constraints are selected from the group of lookup errors, query errors, structural accuracies, physiological accuracies, separation structures, retention of traditions, cross-platform compatibility, layout regularity, and scan time.

10. The apparatus of claim 4, wherein the lookup error is at least 1/50 and the query error is at least 1/5.

11. The apparatus of claim 4, wherein the lookup error is in a range between 1/50 and 1/100 and the query error is in a range between 1/5 and 1/10.

12. The apparatus of claim 4, wherein the lookup error is in a range between 1/100 and 1/200 and the query error is in a range between 1/10 and 1/20.

13. The apparatus of claim 2, wherein the lookup error is less than 1/200.

14. The apparatus of claim 3, wherein the query error is less than 1/20.

15. The apparatus of claim 1 wherein said plurality of keys are 13 keys, said plurality of characters are letters a through z, and each said key is associated with two of said characters.

16. The apparatus of claim 15 including a mode key wherein a first one of the two of the plurality of characters associated with one of the plurality of keys is entered when the mode key is activated while the one of the plurality of keys is activated and a second one of the two of the plurality of characters associated with the one of the plurality of keys is entered when the one of the plurality of keys is activated when the mode key is not activated.

17. The apparatus of claim 1, wherein said ambiguous codes are optimized for a plurality of languages.

18. The apparatus of claim 1, wherein said disambiguation encoding comprises a dictionary of words including at least one uncommon word, wherein said at least one uncommon word is removed from said dictionary.

19. The apparatus of claim 18, wherein said disambiguation code selects among at least two words formed by activating ones of said plurality of keys, each of said at least two words having a probability of occurrence, wherein a first of said at least two words is selected when said probability of occurrence of said first of said at least two words is greater than 100 times said probability of occurrence of a second of said at least two words.

20. The apparatus of claim 1, wherein the plurality of keys are associated with a standard telephone keypad having 8 character keys and 4 non-character keys.

21. The apparatus of claim 20 wherein at least one of said non-character keys has a punctuation character associated therewith.

22. The apparatus of claim 21, wherein the punctuation character is a period or a comma.

23. The apparatus of claim 20 wherein at least one of said non-character keys has an editing function associated therewith.

24. The apparatus of claim 23, wherein the editing function is space and backspace.

25. The apparatus of claim 1, wherein said disambiguation code resides in said keyboard.

26. The apparatus of claim 1, wherein said keyboard is in communication with a remote computer, said disambiguation encoding residing in said remote computer.

27. The apparatus of claim 15, comprising a second 13 key next to the plurality of keys and a mode key, wherein when the mode key is actuated, one of the two of the plurality of characters each associated with the plurality of keys is associated with a respective one of the second 13 keys.

28. The apparatus of claim 27, wherein the second 13 keys are symmetrically arranged next to the plurality of keys, wherein one of the two characters of the plurality of characters each associated with the plurality of keys is associated with a symmetrical one of the second 13 keys, respectively.

29. The apparatus of claim 27, wherein the plurality of keys and the second 13 keys are arranged in three rows, a first and second row of the three rows containing 5 keys and a third row of the three rows containing 3 keys.

30. The apparatus of claim 27, wherein the plurality of characters associated with the plurality of keys have a first aggregate probability of occurrence and the plurality of characters associated with the second 13 keys have a second aggregate probability of occurrence, wherein the first aggregate probability of occurrence is greater than the second aggregate probability of occurrence.

31. The apparatus of claim 27, wherein the plurality of characters associated with the plurality of keys have a first aggregate probability of occurrence and the plurality of characters associated with the second 13 keys have a second aggregate probability of occurrence, wherein the first aggregate probability of occurrence is less than the second aggregate probability of occurrence.

32. The apparatus of claim 27, wherein the plurality of characters associated with the plurality of keys have a first aggregate probability of occurrence and the plurality of characters associated with the second 13 keys have a second aggregate probability of occurrence, wherein the first aggregate probability of occurrence is equal to the second aggregate probability of occurrence.

33. The device of claim 1, comprising a disambiguation input key, wherein one of said plurality of characters is output in accordance with said disambiguation input key when said disambiguation input key is actuated.

34. The apparatus of claim 1, wherein said plurality of keys are associated with a numeric keyboard.

35. An apparatus for touch-key entry of text on a handheld device, comprising:

a keyboard having at least 8 keys, each of the at least 8 keys being associated with a plurality of characters, one of the plurality of characters being designated as a shifted character for each of the at least 8 keys, remaining characters of the plurality of characters being designated as non-shifted characters for each of the at least 8 keys;

a shift key; and

ambiguous code elimination;

the method comprises the following steps: inputting the shifted character associated with the one of the at least 8 keys when the one of the at least 8 keys is actuated while the shift key is actuated, and inputting one of the remaining characters of the plurality of characters associated with the one of the at least 8 keys according to the disambiguation code when the one of the 8 keys is actuated while the shift key is not actuated.

36. The apparatus of claim 35 wherein the characters are associated with the at least 8 keys according to a standard ambiguous code.

37. The apparatus of claim 36, wherein the shift characters comprise c, e, h, l, n, s, and t.

38. The apparatus of claim 37, wherein the shift key comprises x.

39. The apparatus of claim 37, wherein the shift key comprises y.

40. The apparatus of claim 35, wherein said disambiguation code is a word-based disambiguation code.

41. The apparatus of claim 35, wherein said ambiguous code is a block-based ambiguous code.

42. The apparatus of claim 35, wherein said disambiguation encoding comprises using syntax and semantics.

43. The apparatus of claim 35, wherein each of the plurality of characters is displayed on a respective one of the at least 8 keys.

44. The apparatus of claim 43, wherein said shift key is displayed in one font and said non-shift key is displayed in a second font.

45. The apparatus of claim 35 wherein said keypad is a standard telephone keypad having 8 character keys and 4 non-character keys, said at least 8 keys being said 8 character keys.

46. The apparatus of claim 45 wherein said shift key is one of said non-character keys.

47. The apparatus of claim 45 wherein at least one of said non-character keys is associated with a punctuation character.

48. The apparatus of claim 47, wherein the punctuation characters are periods and commas.

49. The apparatus of claim 45 wherein at least one of said non-character keys is associated with an editing function.

50. The apparatus of claim 49 wherein the editing functions are space and backspace.

51. The apparatus of claim 35, wherein said disambiguation code resides in said keyboard.

52. The apparatus of claim 35, wherein said keyboard is in communication with a remote computer, said disambiguation encoding residing in said remote computer.

53. The apparatus of claim 35, wherein said disambiguation encoding comprises a dictionary of words including at least one uncommon word, wherein said at least one uncommon word is removed from said dictionary.

54. The apparatus of claim 53, wherein said disambiguation code selects among at least two words formed by activating ones of said plurality of keys, each of said at least two words having a probability of occurrence, wherein a first of said at least two words is selected when said probability of occurrence of said first of said at least two words is greater than 100 times said probability of occurrence of a second of said at least two words.

55. The device of claim 35, comprising an ambiguous input key, wherein one of the plurality of characters is input based on the ambiguous input key when the ambiguous input key is activated.

56. An apparatus for touch-key entry of text on a handheld device, comprising:

a plurality of keys;

a plurality of characters;

a multilevel ambiguous code, said character being associated with said plurality of keys according to said multilevel ambiguous code; and

ambiguous code elimination;

the method comprises the following steps: inputting one of said plurality of characters according to said disambiguation code when one of said keys is actuated,

the ambiguous code is characterized by being touch-typable.

57. The apparatus of claim 56, wherein said plurality of ambiguous codes comprise a first level of ambiguous codes and a second level of ambiguous codes, wherein in said first level of ambiguous codes said plurality of characters are divided into a first plurality of groups, each of said first plurality of groups being associated with one of said plurality of keys, wherein in said second level of ambiguous codes each of said first plurality of groups is divided into a second plurality of groups, each of said second plurality of groups being associated with one of said plurality of keys.

58. The apparatus of claim 56 wherein said plurality of keys are 4 keys.

59. The apparatus of claim 56 wherein said plurality of characters are letters a through z.

60. The apparatus of claim 57 wherein said plurality of ambiguous codes are optimized for lookup errors and query errors, said first ambiguous code is optimized for alphabetic ordering, and said second ambiguous code is optimized for alphabetic ordering and separation equality.

61. The apparatus of claim 56, wherein said multi-level ambiguous encoding is optimized for lookup errors.

62. The apparatus of claim 56, wherein said multi-level ambiguous encoding is optimized for query errors.

63. The apparatus of claim 56, wherein said multi-level ambiguous code is optimized for touch-typability.

64. The apparatus of claim 63, wherein said multi-level ambiguous code is strongly touch typable.

65. The apparatus of claim 57, wherein said first level ambiguous encoding is optimized for structural accuracy.

66. The apparatus of claim 57 wherein said first level ambiguous code is optimized for alphabetic ordering.

67. The device of claim 57, wherein said second level ambiguous encoding is optimized for equal separation.

68. The device of claim 57, wherein said second level ambiguous encoding is optimized for structural accuracy.

69. The apparatus of claim 57 wherein said second level ambiguous code is optimized for alphabetic ordering.

70. The apparatus of claim 56 wherein each of said plurality of keys has a display portion, each of said display portions displaying said plurality of characters according to said multilevel ambiguous code.

71. The apparatus of claim 56, wherein a respective most likely character is displayed on each of said plurality of keys, said most likely character being displayed at a first position on said respective key.

72. An apparatus as recited in claim 71, wherein the most likely character is displayed in a visually apparent manner.

73. The apparatus of claim 56, wherein said disambiguation encoding comprises a dictionary of words including at least one uncommon word, wherein said at least one uncommon word is removed from said field.

74. The apparatus of claim 73, wherein said disambiguation encoding selects from at least two words formed by activating ones of said plurality of keys, said at least two words each having a probability of occurrence, wherein a first of said at least two words is selected when the probability of occurrence of said first of said at least two words is greater than 100 times the probability of occurrence of a second of said at least two words.

75. The apparatus of claim 56, wherein said disambiguation code resides in said keyboard.

76. The apparatus of claim 56, wherein said keyboard is in communication with a remote computer, said disambiguation encoding residing in said remote computer.

77. The device of claim 56, comprising a disambiguation entry key, wherein one of said plurality of characters is entered based on said disambiguation entry key when said disambiguation entry key is actuated.