HK1118970A - Typability optimized ambiguous keyboards with reduced distortion - Google Patents

Description

Reduced distortion typewriteability optimized ambiguous keyboard

Technical Field

The present invention relates generally to an ambiguous keyboard-based computerized text input system, and more particularly to a typeability (typability) optimized ambiguous keyboard with reduced morphing (distortion).

Background

Introduction to the word

The first reaction to the change is a rejection. To improve the usability of the keyboard, it may be necessary to change its appearance. However, changing the keyboard to a familiar design may make the keyboard appear less usable at first sight. The perceived usability and the actual usability are not consistent. Thus, there is a long felt but unexpressed need for: keyboards were designed which, although novel, were considered usable because they were similar to commonly available products. While a similar anxiety is caused when introducing many new technologies, the present invention presents a solution to the problem in the specific field of ambiguous keyboards. Disclosed herein is an ambiguous keyboard design that is novel over conventional designs in that it has improved typeability, but with substantially minimized distortion over conventional designs, which is easy to get and practice for new users.

To minimize the deformation, the deformation must be properly defined, measured and controlled. Also, to maximize typeability, typeability must be properly defined, measured, and controlled. To achieve the object of the invention, new measures for both morphing and typeability are introduced. It is shown how these new and prior art measures can be used to synergistically combine both distortion minimization and typeability maximization. This gives an unexpected result of making the device attractive to both novice and advanced users.

The present invention introduces a novel class of devices with acceptable layout variations and acceptable typeability, both of which are important enough to require a compromise between the two. Prior art methods seek to optimize only for one set of constraints or another, and thus only consider certain aspects of layout distortion or typeability. Prior to the advent of the relied upon us patent 6885317 to Gutowitz (incorporated herein by reference, hereinafter Gutowitz' 317), no mention was made in the literature: layout morphing and typeability are linked while being much less optimized as taught by the present invention.

The present invention teaches how to construct a device that synergistically reinforces the teaching of maximizing typeability and the teaching of minimizing distortion. Measuring or minimizing deformation is particularly highly non-obvious because deformation is a psychological property rather than a physical property. The initial impression of the device (the prospect of usability that the design conveys through its appearance) is at least as important to the commercial success of the device as the actual usability of the device when in use. Designs seeking to enhance typeability without limiting distortion generally do not succeed. For example, although the Dvorak keyboard (fig. 3C) was greatly blown and claimed to have improved typeability over the dominant qwerty keyboard, it was not successful. This failure may be traced back to the fact that: dvorak does not attempt to smooth the cleft between his keyboard and the conventional keyboard.

Since designers of prior art keyboards either blindly adhere to the conventions or aggressively change conventions, nothing has so far taught how or how we combine typeability optimization and warping constraints. While prior art designers have focused on initial product adoption or performance for skilled users, a product must focus on both aspects if it is to be truly successful. The present invention teaches how to seek to improve the commercial success of keyboards in a systematic manner.

Although we are concerned with the appearance of the device, our findings are in the engineering field and not in the aesthetic field. We seek to design to feel comfortable and familiar, rather than to feel beautiful. To achieve these engineering goals, several novel measures are introduced that capture the intuitive meaning of "deformation" when computing the physical properties of a layout. With these measures, a search can be made in the space of alternative layouts for a layout that satisfies the design constraints.

The earliest period of concern in ambiguous keyboard design to date was the discovery phase (U.S. patent application 10/415,031 to Gutowitz and Jones). During the discovery period, the user actually manipulates the device. During the previous discovery period (the period when one assumes that one would like to use the device), the appearance of the device is critical. The pre-discovery phase is a major concern of the present invention.

Definitions and basic concepts

This section collects definitions of words and concepts that will be used in the following detailed description.

The language. Given a set of symbols, one can construct sequences of symbols and assign probabilities to those sequences. The set of symbols, the sequences of symbols and the probabilities assigned to the sequences will be referred to herein as the language. For clarity of discussion and without limiting the scope of the invention, we will refer to the language as written natural language (e.g., English), and although we may refer to the symbol as "letters" or "punctuation" for concreteness, one of ordinary skill in the art will appreciate that the symbol discussed herein may be a writing of any discrete unit, including standard symbols (e.g., Chinese ideograms) or inventive symbols (e.g., the name of an artist previously referred to as Prince).

And (4) carrying out polysemous coding. Ambiguous codes are well known in the art. There are 12 keys on a standard telephone keypad used in the united states, 10 of which are encoded with a number and several of which (usually 8) are also encoded with 3 or 4 letters arranged alphabetically. These allocations produce ambiguous codes, which we call Standard Ambiguous Codes (SAC). The code is abc def ghi jkl mno pqrs tuv wxyz.

A method of disambiguating. Since several letters are encoded on each key in an ambiguous code, some disambiguation method must be used to determine which of the several letters is intended by the user. The disambiguation method is typically software that predicts the letter sequence the user wants based on the user's previous input and a database of linguistic information.

A conventional keyboard. There are mainly three widely used standard keyboards for the latin alphabet: qwerty keyboards and their close variants, and 12-key telephone keyboards with the standard ambiguous codes described above. Other written letters have other keyboards, and it should be understood that any of the devices or methods described herein are equally applicable to those keyboards for other written letters.

And (6) layout. A layout is an assignment of letters to keys, where the keys are in some spatial arrangement. The terms keyboard and layout may be used interchangeably when not confusing.

And (5) layout deformation. In this disclosure, we are concerned with a pair of keyboards: conventional keyboards and morphed keyboards that are derived from conventional keyboards by moving some letters away from their positions in the conventional keyboard. Layout variations are the differences between the conventional keyboard and the derived keyboard. There are two main categories of layout variations: order distortion and division distortion.

And (5) deforming the sequence. The order of layout is the order in which a reader of a language in which letters are typed by a keyboard reads the labels of the keys, e.g., english is typed with latin letters by a qwerty keyboard, and the keyboard is read from left to right, top to bottom: qwertyiupasdfgh …. Order distortion is the displacement of a letter from its regular position in that order.

And (4) dividing. The division of the integer n is a set of integers, the sum of the elements in the set being equal to n. In general, a given integer allows for multiple partitions, e.g., integer 5 has a partition of 3: 2, but also has a partition of 2: 1. Algorithms for generating all the divisions of integers are well known to those skilled in the art. There are various characteristics of the divisions that are relevant to the present disclosure, some of which are defined immediately below.

As homogeneous as possible. Most prior art coding uses a partitioning that is as uniform as possible (even-as-possible). I.e. a division in which the number of keys associated with the number of letters to be coded is given as much as possible, the number of letters per key being the same. As homogeneous as possible may be abbreviated EAP.

The line is deformed. Most conventional keyboards include keys organized in a regular (usually honeycomb-like) array with identifiable rows and columns. A new layout has row deformations if a letter is moved away from its regular row in the new layout. Column deformations are defined in the same way.

And (3) a range. The extent of the division is a generalization of the properties to be as uniform as possible. The irregularity of the division is defined as the difference between the minimum number and the maximum number of letters assigned to an arbitrary key. If the conventional keyboard is a non-ambiguous keyboard with one letter per key, the less irregularity the warped keyboard intuitively perceives, the less warped the keyboard.

A dominant category. The dominant category of division for letters on a key is the maximum number of keys with the same number of letters. Thus, the dominant category of division (4, 3, 3, 1) for letters on a key is two keys each having 3 letters. Intuitively, the larger the dominant category relative to the total number of keys in the division, the more regular the keyboard. Both divisions may have the same range but a different number of keys in the dominant category.

The posture is deformed. Layout variations can be classified for the following conditions: moving the letters from their positions in a conventional keyboard to whether and to what extent the warped keyboard changes the gesture used to type the letters. For example, swapping the letters q and a on a qwerty keyboard does not have an effect on which finger is used to type q or a, and thus, while the swap does change the distance that the finger must move in order to type the letter, the swap is equi-fingered. In both the qwerty keyboard and the morphed keyboard, q and a are both typed by the touch typist with the left little finger.

Typeability. Typeability refers to the work or time required to enter text. A common measure of the operation of an ambiguous keyboard is kspc (average number of keystrokes per character). The amount of time required to enter text cannot be simply related to kspc. In order to input text, various processes other than pressing a key must be performed, and these processes take time. For example, if a word-based disambiguation method is used and more than one word corresponds to the keystroke sequence used to input the desired word, time is required to check and select the desired word from the possible candidates.

Drum beating effect. The drumlole effect (drumroll effect) is a typeability constraint related to the time required to enter text. In general, not all keystrokes take the same amount of time. For example, if each of a pair of letters in a sequence is typed with a different finger, the sequence may be entered faster than if they were typed using the same finger. While the first finger is inputting the first letter, the second finger may be moved to a position where the second letter is to be input. Thus, the first keystroke and the second keystroke overlap in time. This overlap is referred to as the drum beating effect.

Fitts Law. Fitts's law is a mathematical model used in typing studies to estimate the time required to strike a key based on the size of the keys and the distance between the keys. The longer the distance, the longer the time, and the larger the key, the shorter the time.

Steric hindrance. The technical term is used from the chemical field. It refers to the obstruction between freely moving objects whose movement is obstructed when the objects are close to each other due to the fact that the objects occupy space. Steric hindrance must be considered when the size of the key is small compared to the size of the finger or thumb used to strike the key. Steric effects can alter the results of both drumming and Fitts' Law analysis.

And (4) an interaction mechanism. The interaction mechanism is the physical means by which a user interacts with the keyboard. For example, a telephone keypad is typically struck with one finger or one thumb or two thumbs. Which interaction mechanism to use may depend on many factors, depending on the user's experience and/or other activities the user is engaged in when making text input, e.g., holding a coffee cup in one hand may discourage the user from using the two-thumb interaction mechanism she would otherwise like. Some typeable measurements rely on an interaction mechanism, while others do not.

Disambiguating the software. When there is more than one letter on a key, some means is required to select the letter the user wants at any given time. The selection may be mechanical (e.g., one keystroke is the first letter and two keystrokes is the second letter, …), or the selection may be determined according to an algorithm that guesses what is desired based on context and language statistics. Such software is called disambiguation software.

Next function/key. If the currently displayed word is incorrect, the word-based disambiguation system uses a "Next" function to allow the user to change the displayed word, and if the currently displayed letter is incorrect, the character-based disambiguation system uses a "Next" function to allow the user to change the displayed letter. These functions are collectively referred to as a "Next" function, and the key that performs the function is referred to as a "Next" key.

A typeability optimized keyboard with minimal distortion. If the keyboard with given value deformation is one of the best keyboards with respect to the typeability constraint, the keyboard is considered optimized with respect to the typeability constraint and the keyboard has substantially the given value deformation. For example, let the typeability constraint be the lookup error rate, and let the distortion measure be the number of pairwise swaps that map the distorted keyboard to the qwerty keyboard. If the constraint on warping is 5 pairwise swaps, then the optimized keyboard with warping constraint 5 is one of all keyboards warped to 5 or less with the best lookup error rate.

Drawings

FIG. 1) stages of product adoption.

Fig. 2) summary of some related art.

FIG. 3) representations of the Dhiatensinor keyboard and the Dvorak keyboard.

FIG. 4) a representation of qwerty, azerty, 5 columns of qwerty, and the traditional keyboard of the slavic (Cyrillic) language.

Fig. 5) a layout of reserve order warp and no reserve order warp of Gutowitz' 317.

Fig. 6) Matias' U.S. patent No. 5,288,158.

FIG. 7) a block diagram of an ambiguous keyboard based typeable device.

FIG. 8) a chart of exemplary typeability constraints.

Fig. 9) touch typability area and number of valid keys as defined by Gutowitz' 317.

Fig. 10) an exemplary character-based disambiguation results of U.S. patent No. 6,219,731 to Gutowitz.

FIG. 11) a chart of exemplary appearance distortion constraints associated with partitioning.

Fig. 12) a chart of exemplary appearance distortion constraints with respect to a sequence.

Fig. 13) legend for the design of quantitative distortion measures associated with the partitions.

Fig. 14) follows the Gutowitz' 317 as uniform qwerty-like keyboard layout as possible for various column numbers.

FIG. 15) a diagram of exemplary gesture deformation constraints.

FIG. 16) a flow chart of a method for producing a reduced distortion typeability optimized keyboard.

Fig. 17) illustrates a summary chart of an embodiment of the tradeoffs of typeability and distortion.

FIG. 18) is a flow chart of an exemplary method of generating a reduced distortion actual typeability optimized keyboard for a telephone keyboard.

Fig. 19) the number of valid keys of the optimal layout with a given value of layout range and out of order distortion.

Fig. 20) layout corresponding to the dots of fig. 19.

Fig. 21) distribution of the number of valid keys as a function of the number of order variations.

Fig. 22) summary of the results of applying the method of fig. 18.

Fig. 23) applies the exemplary best results of the method of fig. 18.

Fig. 24) results of applying the method of fig. 18 to a 5-column qwerty type keyboard.

Fig. 25) a 5-column qwerty-type keyboard with a range of order variations.

Fig. 26) a diagram of an exemplary navigation keyboard.

Fig. 27) the alphabetical order preserved layout of the navigation keyboard.

Fig. 28) the layout of the navigation keyboard that preserves the qwerty keyboard order and the two-thumb gesture.

Fig. 29) conceptual difference layout of the navigation keyboard.

Fig. 30) the layout of the navigation keypad with the telephone keypad row preserved.

Fig. 31) legend of steric hindrance caused by large ratio of thumb size/bond size.

Fig. 32) applies the drum-beating constraint to evaluate the two-key layout.

Fig. 33) an embodiment of drum beating optimization for steric hindrance by sign multiplication.

FIG. 34) the gesture-preserved qwerty class layout for the steering wheel.

FIG. 35) typeability distribution of keyboard typeability optimized for two different interaction mechanisms simultaneously.

FIG. 36) an example layout optimized for two interaction mechanisms simultaneously.

Fig. 37) a flowchart of predictive compensation for distortion.

Fig. 38) a first exemplary embodiment of a chord (chord)/polysemous code for a gaming apparatus.

Fig. 39) a second exemplary embodiment of a chord/polysemous code for a gaming apparatus.

Fig. 40) exemplary embodiments of a chord/polysemous coding layout with optimized typeability and minimized appearance distortion.

Fig. 41) illustrates a table of the synergistic effect of the division distortion and the order distortion.

Fig. 42) a portion of an exemplary embodiment of a variable layout keyboard family.

FIG. 43) a full-size member of a variable layout keyboard family.

Fig. 44) a comparison of a prior art data device and a data device according to the present invention.

FIG. 45) an exemplary embodiment of a context-based disambiguation keyboard.

Fig. 46) an exemplary embodiment of a link/unlink mechanism.

Disclosure of Invention

The present disclosure begins by building a framework in terms of product adoption phases. The present disclosure, then, explains the meaning of the deformation and the typeability by means of many embodiments, and shows how to measure both of them.

A number of non-limiting embodiments are shown by way of example to illustrate the scope of the invention. The scope is not limited by the kind of typability or morphing discussed, and the particular clustering of typability constraints and morphing constraints used in various embodiments is to illustrate how these separate concepts so far are synergistically combined. More than one typability and more than one type of morphism may be combined, and may also be combined with other types of morphism and typability not discussed herein. The disclosed principles function in a very general arrangement allowing for many variations as will be appreciated by those skilled in the art. The non-limiting embodiments discussed herein are meant to be illustrative only, with the true scope of the invention being indicated by the following claims.

Object of the Invention

One objective is to create an ambiguous keyboard that is optimized for more than one stage of the product adoption process.

One objective is to optimize the keyboard with respect to typeability constraints including, but not limited to: lookup error, qwerty error, number of valid keys, number of key strokes per character, drum beating probability, valid drum beating probability, Fitts' rule, throughput, robustness, and language generality.

One objective is to optimize the keyboard with respect to partition-related appearance deformation constraints, including but not limited to: as uniform as possible, maximum or minimum number of letters per key, range, dominant class, bilateral symmetry, up-down symmetry, and monotonicity.

One objective is to optimize the keyboard with respect to order-dependent appearance distortion constraints, including but not limited to: a read order, a row limited read order, a column limited read order, an external weighting, a row limited letter movement, a column limited letter movement, a distance limited letter movement, a number of letter shifts, and a number of letter swaps.

One objective is to associate the apparent deformation with a quantifiable mathematical model suitable for use in an optimization method.

One objective is to optimize the keyboard with respect to pose deformation constraints including, but not limited to: the same finger, symmetrical fingers, same hand, adjacent fingers, and same gesture category.

One object is to show how to create and use a reduced-distortion typeability optimized ambiguous keyboard.

Another object is to present an appearance-morphing optimized ambiguous keyboard optimized for typewritability.

Another object is to present a gesture-morphing optimized ambiguous keyboard optimized for typeability.

Another object is to provide a morphically optimized ambiguous keyboard optimized for drumming typewritability.

Another object is to present a layout based on conceptual differences.

It is an object to present a keyboard that is optimized with respect to finger blockage.

Another object is to present an ambiguous keyboard optimized for more than one typeability metric.

Another object is to present a practical solution to map a conventional keyboard to a telephone keyboard while optimizing typeability and reducing distortion.

Another object is to present an ambiguous keyboard optimized for more than one distortion measure.

One object is to present an ambiguous keyboard with optimized gesture deformation that fits a gripping object such as a steering wheel or a handle bar.

Another object is to present an ambiguous keyboard with optimized gesture deformation suitable for navigation keyboards.

Another object is to present an ambiguous keyboard for an alphabetically based navigation keyboard.

Another object is to present an ambiguous keyboard for a navigation keyboard based on alphabetical order and compatible with the telephone keyboard row.

Another object is to give an appearance-morphing optimized ambiguous keyboard optimized for typeability compatible with a keyboard comprising three rows and 1 to 9 columns.

It is an object to present an appearance-morphing optimized ambiguous keyboard optimized for typeability and compatible with a telephone keyboard.

It is an object to present a morphed optimized keyboard having two letter keys.

One object is to present a layout morphing keyboard that is easy to interpret and remember.

It is an object to present a keyboard that is optimized for more than one interaction mechanism.

Further objects will become apparent from the detailed description of the invention which follows.

Detailed Description

Introduction to the word

FIG. 1 gives an overview of the invention, showing how aspects of the invention relate to the maturity stages of a user's product adoption process.

Is met. In the encounter phase, the user has not used the device, but just has seen it (possibly in a photograph). The only experience a user may have with using the device is his or her mental prediction of how the device is used. We refer to this psychological prediction as initial perceptual usability. The initial perceived availability will be based on the actual experience that the user already has for similar devices. One of the findings on which the present invention is based is: initial perceived usability may be maximized when layout distortion from a conventional layout is minimized.

And (5) finding. In the discovery phase, the user starts operating the device and tries to use it to enter text. Studies have shown that users typically only perform a few initial experiments of entering text, and that users may abandon the device if these initial experiments are not seen, i.e. if the device appears to be difficult to use, does not give the intended result, or "feels bad". Therefore, it is critical that the disambiguation software cannot generate too many errors, which would otherwise confuse the user at this stage. The number of errors generated by disambiguation software is partially layout dependent. Given a particular disambiguation method, the layout may be modified to reduce the number of errors. One aspect of the present invention is to solve the following design problems that arise: modifications to the layout to reduce disambiguation errors typically reduce the initial perceived usability, as these modifications distort the keyboard layout from its normal form. Thus, optimization for success in the discovery phase may conflict with optimization for success in the encounter phase.

And (5) learning. In the learning phase, the user who has decided to use the device starts to master, seeking speed and accuracy of text input by continuing to exercise. Good disambiguation (of importance in the discovery phase in the first place) remains important. By contrast, the initial perceived usability has been relatively diminished, since the user's present perception is based on the actual use of the device. The effects of conventional designs remain as they continue to work with the user's deeply rooted athletic posture through long term use of conventional designs. As with learning to ride a bicycle subject to an already learned pattern of walking motion, any retention of posture from a conventional keyboard to a novel keyboard based on a conventional keyboard will accelerate learning of the novel keyboard. Thus, another aspect of the present invention provides a keyboard that minimizes the gesture distortion used to operate a conventional keyboard and is also optimized with respect to disambiguation mechanisms.

And (4) skillfully preparing. In the proficiency phase, not only the initial perceived usability has been replaced by the actual experience of using the device, but also the conventional gestures have been modified or replaced by gestures taken against the new keyboard. The user of the new keyboard can form an interaction mechanism with the device that has little to no relation to the conventional interaction mechanism on which it is based. Another aspect of the invention performs skilled interaction mechanism optimization in a manner that minimizes the splitting of optimizations designed to improve user experience at an earlier stage of development.

The encounter phase, discovery phase, learning phase and proficiency phase are similar to the romantic phases (roughly beginners, moods, love, marrying). This analogy is relevant because users form a very rudimentary model of interacting with their keyboards and also choose which keyboards to use based on criteria that are quite different from the criteria critical to success at the later stages of the romantic relationship.

Taught by the illustrated embodiments and claimed in the appended claims is a set of devices that synergistically combines optimizations made for more than one level of keyboard adoption. The present disclosure seeks to make it apparent to one of ordinary skill in the art how to balance optimization for one phase with optimization for another phase, resulting in a keyboard that is both likely to be employed and that will work efficiently once employed.

It will be appreciated that it will be easier to perform such optimization for only one phase. The keyboard that is optimal for each stage can be selected. However, learning a keyboard refers to learning the motor reflexes of quickly entering symbols, and if the keyboard is changed halfway, these gestures must be relearned. Furthermore, typical hardware keyboards do not allow easy rearrangement of letters with respect to key assignments. Thus, the present invention solves a difficult and heretofore unseen problem.

Prior Art

Turning to FIG. 2, we find a diagram of a selected related art keyboard.

The qwerty keyboard (fig. 4A) is a prototype of a conventional keyboard layout. It is well established as a convention in the english world, and other latin alphabet languages often use a conventional keyboard that is a close variant of qwerty. An example, an azerty keyboard for use in france, is shown in fig. 4B. While these other keyboards may be considered variations of the qwerty keyboard, they are not ambiguous keyboards and they are not optimized for typeability. Other conventional keyboards exist for other letters, such as the keyboard of fig. 4D for the letters of slav.

The Dhiatensinor keyboard (FIGS. 3A and 3B) is relevant because it is an early example of a keyboard optimized for a two-finger interaction mechanism. The letters are placed in probabilistic order from the center outward and from the bottom row to the top row. The Dhiatensor keyboard is not an ambiguous keyboard and it is not a variation of the conventional keyboard. In fact, the keyboard was designed before there was a well established convention for typewriter keyboard layout.

The Dvorak keyboard (fig. 3C) was optimized for an 8-finger interaction mechanism. It seeks to minimize the distance that the finger must travel to type the most commonly used letters. The Dvorak keyboard is not an ambiguous keyboard and it is not a minimal variation. Although qwerty was well-customized to the convention at the time of the invention of the Dvorak keyboard, Dvorak did not attempt to retain any portion of the convention in its design.

The Matias' half qwerty keyboard of FIG. 6 (U.S. Pat. No. 5,288,158) is a keyboard with limited gesture deformation. It attempts to preserve the typing posture from the qwerty keyboard by "folding" the qwerty keyboard in half so that letters typed with a given finger on the qwerty keyboard are typed with the same finger on the half qwerty keyboard (although possibly with a different hand). The half qwerty keyboard is not an ambiguous keyboard and it is not optimized for typeability.

U.S. patent application Gutowitz 09/856,863, which was incorporated herein by reference as of the filing date of the present application, will be referred to hereinafter as Gutowitz' 317. Gutowitz' 317 provides background for a number of new inventive concepts presented herein. This publication introduces a division warped keyboard and an order warped keyboard of the qwerty class, explores the advantages of a layout as uniform and not as uniform as possible, and provides focus on a layout of two letters per key. Some example embodiments from Gutowitz' 317 are shown in fig. 5. Fig. 5A shows a partitioned variant of a conventional alphabetical layout for a telephone keypad. FIG. 5B shows a qwerty type layout on 7 columns with monotonically decreasing number of keys assigned letters per row, with a partitioning variation for optimizing typeability. FIG. 5C shows a qwerty class layout on a 7 column with a partition warp and an order warp. The number of order deformations (8) shown in this figure is quite large compared to what is considered in this disclosure as a "near qwerty" layout. The layout also does not comply with other ordering constraints (such as keyboard name constraints) which will be discussed in detail below.

The 5-column qwerty keyboard of fig. 4C is a qwerty-type keyboard that is as uniform as possible. This arrangement is used in a non-ambiguous manner by us patents 5661476 and 6295052. As just mentioned, the use of ambiguous codes for qwerty type keyboards (including as uniform and not as uniform as possible) was first proposed by Gutowitz' 317 and used In its 7100x phone by Research In Motion corporation In a commercial environment. This as uniform layout as possible represents a tight partitioning constraint, and therefore leaves little margin for trading off with the typeability constraint. As will be discussed below, the 5-column design achieves a layout that is much higher in typeability than the most uniform possible layout of fig. 4C.

Qwerty-like ambiguous keyboard as uniform as possible and distortion of appearance

Gutowitz' 317 includes a ambiguous keyboard that is as uniform and not as uniform as possible. As uniform as possible is the basis on which the apparent deformation can be measured. Intuitively, ambiguous keyboards that are as uniform as possible have a relatively low distortion of appearance, since the conventional keyboards on which they are based are trivially as uniform as possible (since each key has exactly one letter). To be a qwerty class, the reduced keyboard should preferably: a) have the same letters as qwerty in each row, and b) the number of keys with letters monotonically decreases as the rows increase from top to bottom. Some examples of varying and monotonically decreasing column counts are shown in fig. 14 as uniform a keyboard as possible.

Since there is one or very few as uniform layouts as possible for a given number and arrangement of keys, it is trivial to optimize for typeability over the set of as uniform layouts as possible. A difficult problem recognized and subsequently addressed by the present invention is to limit non-trivial levels of morphing and then optimize typewritability while respecting the limit. The consumer is expected to accept the keyboard as long as the morphed keyboard remains with little variation from the conventional keyboard. Even if the variation remains small, the design should maximize typeability. As can be seen from fig. 14, the first most uniform layout possible, which just achieves the minimum level of touch typability (a-level touch typability of Gutowitz' 317), is a 4-column layout. It is important to implement touch typability with a 3-column keyboard, as keyboards are extremely common. This problem will be returned to below.

Method

In this section, we will discuss two main features of interest to the present invention: typeability and deformability.

Typewriting ability

Typeability refers to the property that affects text throughput when text is entered using an ambiguous keyboard. How many keystrokes are required for each character? How many errors the system generates? How does the system respond when the user has an error? The typeability feature results from the interaction of the keyboard with the disambiguating software. Recall that the ambiguous code based typeable device has three main components. Referring to fig. 7, we see a block diagram depicting these components. The disambiguation keyboard 701 sends keystrokes to the disambiguation software 702, and the disambiguation software 702 decodes the keystroke sequences as good as possible into text and sends the text to the output 703.

There are many factors that affect the text throughput through the device depicted in fig. 7. Some of these factors are listed in the chart of fig. 8. Some factors are related only to the keyboard, such as the difficulty of pressing a key, and some factors are related only to the disambiguation system, such as the number of words in the dictionary-based system. We are often concerned with features, such as lookup errors, caused by the interaction of the keyboard with the disambiguation system. The lookup error is the rate at which the word-based disambiguation system guesses the wrong word (the word that the user does not want, but that has the same sequence of keystrokes as the word the user wants). This feature depends on both the disambiguation system and the keyboard layout.

To help understand how keyboard layouts are related to typeability, we will quickly review character-based disambiguation methods and word-based disambiguation methods, as well as measures of their typeability. This material is contained in more detail in U.S. patent No. 6,219,731 to Gutowitz and Gutowitz' 317, both of which are incorporated herein by reference and dependent thereon. In particular, Gutowitz' 317 defines several measures of typeability for a word-based disambiguation system: significant search errors, query errors, number of valid keys, and A, B, C-level touch typability. The performance of a disambiguation system with an effective number of n keys is the same as the best performance that can be achieved on a keyboard with n letter keys, if the letters can be arbitrarily assigned to the keys to maximize typeability. In all cases we will consider here, letters cannot be assigned arbitrarily to keys. In fact, we focus here on bringing the layout under strict constraints as close as possible to a given layout. Thus, the effective number of keys for the layout we will discuss will be much less than the number of alphabetic keys they own. The relationship between the number of valid keys taken from Gutowitz' 317 and touch typability at level A, B, C is shown in fig. 9.

A more relevant measure of typeability for character-based predictions is the number of keystrokes per character. In these systems, the user presses a key and then uses the Next key to advance the order of the letters assigned to that key in order of likelihood given the previously defined context of the other entered letters. In Gutowitz' 731, a present figure 10 is presented showing the expected number of key strokes per character as a function of the position of the letter in the word. This is done for two systems: the standard non-predictive multi-tap system, and the predictive character-based disambiguation of Gutowitz' 731, available on essentially all cellular telephones.

Word-based disambiguation and character-based disambiguation are merely aspects of the more general framework of context-based disambiguation, as discussed in gutwitz' 317. The disambiguation of each subtype may have a corresponding typeability metric that is best suited for it. In particular, as noted in Gutowitz' 731, it is apparent even to a person with poor skill in the art to add word or phrase completions to any existing text input method that does not have word or phrase completions. If word completion or any other feature is added to an existing text-based approach, then the quantitative measures described herein also need to be modified to account for the new feature, without departing from the scope of the invention.

1.0.1 measuring and modeling deformation

Throughout, we will use the qwerty keyboard as an example conventional keyboard. It should be understood that the discussion applies equally to any other conventional keyboard. The conventional qwerty keyboard is characterized in that:

1) each key has 1 letter; 2) as the rows change from top to bottom, the number of keys assigned letters monotonically decreases.

The minimally morphed keyboard will have a distribution of letters on the keys as close to this as possible. The most deformed keyboard will have a distribution of letters on the keys as far away from it as possible.

In general, we will consider layouts with different numbers of keys assigned letters in each row. However, to simplify the present exemplary discussion, further limitations are made: each key in the 3 x 3 array has at least one letter assigned to it.

The next step is to assign a numerical measure to the amount of deformation. There are various ways to do this. To be effective, the chosen metric should be a good model of the sensory or interaction constraints to be optimized. Those skilled in the art of mathematical modeling will appreciate that models and phenomena must be distinguished. In the case of a deformation of the appearance, the phenomenon is psychological: how much the reference regular keyboard and the morphed keyboard are perceived to be similar? Those skilled in the art of scientific methods will know how to measure this phenomenon in the laboratory and those skilled in the art of mathematical modeling will know how to construct a mathematical model of the phenomenon. According to the mathematical model, the calculation for performing the minimization of the required deformation can be performed more quickly than by direct psychological studies. Similarly, scientific observations of human interaction with the keyboard, measurements of the anatomy and physiology of the hands, etc. guide those skilled in the art of scientific methods to derive descriptions of gestures used while typing. In fact, there is a large body of literature on this subject. From these experiments and literature, one skilled in the art of mathematical modeling can derive a model of pose deformation. The models and resulting optimized keyboards discussed in this disclosure are non-limiting embodiments selected for their ability to teach one skilled in the art how to create and use keyboards with limited distortion and optimized typeability.

For the purpose of illustration, some variant numerical models that intuitively "look like a qwerty layout as much as possible" will now be considered.

Two metrics are considered:

1) d-morph-the sum of the number of letters on a key minus 1 over all keys.

2) D-morph-the sum of the squares of the letters on the keys over all keys.

Two extreme cases are illustrated in fig. 13. The deformation D of FIG. 13A is 17 according to the measure 1) and 18 according to the measure 2)²+8 × 1 ═ 332. There are 8 other layouts with the same variations, with the largest number of letters on keys 2 to 9, and one letter per key on the other keys. The other extreme in terms of uniformity is shown in fig. 13B, which has 3 letters per key, except that there is one key with two letters. According to measure 1), the deformation of fig. 13B is 8 × 2+1 ═ 17, according to measure 2), the deformation is 8 × 3 (3)²)+2²76. In other words, measure 1) does not distinguish between FIG. 13A and FIG. 13B in terms of distortion; each of fig. 13A and 13B has the same value of deformation (17). However, FIG. 13B is more qwerty-like than FIG. 13A for most people. This indicates that the metric 2) can more correctly express the feeling of similarity to qwerty than the metric 1. The magnitude 2 gives the deflection value (76) of fig. 13B lower than it gives the deflection value (332) of fig. 13A.

Using metric 2), the value of fig. 13C is 78, which is greater than the value 76 of fig. 13B. However, FIG. 13B appears more similar to qwerty than FIG. 13C. The reason for this is that in fig. 13B, there are several letters that are not on the same row as they are in the full qwerty keyboard, whereas in fig. 13C, they are on the same row. This indicates that modifications are to be made to the metric, such as a reduction of letters that are not in the correct row,

wherein L is_keyIs the number of letters on the key, if the letter l does not resemble itIn the same row as in qwerty, g (l) is 1, otherwise it is 0. This will give us values 402, 96 and 76 for fig. 13A, 13B and 13C, respectively. This is a better ranking of these layouts, as it fits our perception of distortion better.

It is again emphasized that the metrics used herein represent exemplary embodiments. It has the following advantages: these keyboards are simply and correctly ordered by their intuitive sensory distortion. Any reasonable (in a realistic sense) measure of deformation may be used where appropriate.

Psychological tests may be performed to determine functional models that are more consistent with human perception than the simple models considered herein. A more accurate model does not change the scope of the invention, but only the numerical values assigned to the keyboard layout. In such psychological testing, various layouts are presented to a large number of subjects many times, and participants are asked to select those layouts from a set of layouts that they consider more similar to qwerty.

In general, we can (at least) distinguish two classes of layout characteristics that may be part of a quantitative model of human perception of similarity: partition related characteristics and order related characteristics. Some exemplary partition-related properties are listed in the figure, and some exemplary order-related properties are listed in figure 12. The split property must deal with the distribution of letters on the keys. And the order-related characteristic relates to the position of the letters in the conventional order of the letters represented in the conventional layout.

Some exemplary constraints will now be reviewed more briefly, which can be applied using the teachings of the present invention to design a useful keyboard. In view of this disclosure, it will be apparent how to apply these and other constraints to optimize typeability while respecting these constraints.

The first set of constraints applies to the apparent deformation. The second set of constraints applies to pose deformation. We will consider various exemplary embodiments showing combinations of these constraints with various interaction mechanisms and typeability metrics.

These varied embodiments are intended to demonstrate that any given set of distortion constraints or typeability measures can be combined in accordance with the teachings of the present invention. These embodiments were chosen to illustrate various aspects of the present invention. In view of this, an intermediate design or a hybrid design should be clearly appreciated by those skilled in the art.

Division deformation

An exemplary partitioning variation is shown in fig. 11. These characteristics are related to the visual balance and coordination of the keyboard. For example, the range of the division, the difference between the maximum and minimum number of letters on the key describes the uniformity characteristic. The advantage of the characteristics related to the partitioning is that they are an easily measurable aspect of the layout. Whether this aspect is important for psychological perception of similarity is a matter of psychological testing. It is important from the standpoint of the present invention that one of ordinary skill in the art can use these and other quantities as a means of developing a mathematical model. The model can then be used as a basis for screening the space of alternative layouts in an attempt to identify the best of those layouts based on the key factors (typeability and distortion) identified herein. In the exemplary embodiments given below, we will consider how some of these quantities can be used to produce a useful keyboard. After considering these examples, one skilled in the art would be able to use other measures, alone or in combination, to select a keyboard with good typeability and appearance characteristics.

Sequence transformation

Order morphing is a change in the order in which symbols are read from the keyboard. To define this, we must formulate the regular reading order of the keyboard. Natural written languages typically have a preferred reading order, which is inherited by the keyboard used to write the language. English is read from left to right, top to bottom, and qwerty keyboards are typically read in the same manner. The name "qwerty" comes from reading the first six letters of the keyboard. The Hebrew keyboard is read from right to left.

With some exceptions. The Dhiatensor keyboard of fig. 3A and 3B is read from left to right, from the bottom row to the top row, which results in the name "Dhiatensor" (the first few letters in this reading order). The standard ambiguous coded "abc" keyboard is read from left to right, from the top row to the bottom row. A given keyboard may allow multiple readings, as indicated by multiple names. The dominant convention for a "qwerty" keyboard is from left to right, from the top row to the bottom row. However, (Neuman, alfred.1964) suggests reading the keyboard from right to left, from the top row to the bottom row, which gives the name "poiuyt". If the proposal becomes conventional, then the letters should be retained in that order, in addition to or instead of "qwerty", in accordance with the teachings of the present invention. The half qwerty keyboard of fig. 6 may be read in qwerty and Neuman order.

FIG. 12 presents a diagram of some exemplary appearance order constraints related to order. Some of these constraints will be used to form the embodiments of the invention below. The respective constraints may be components of a model for quantifying the perceived deformation. For example, studies have shown that if the first and last letters of a word are correct, but the internal letters of the word are changed, there is a high probability that one can still read the word. If the readings for a conventional keyboard have the same characteristics, the model may give higher weight to changes occurring at the boundaries of the key layout than to changes occurring at the center.

Posture deformation

Gesture morphing is important for those who actually use keyboards rather than just looking at them. Anyone trained to touch type on qwerty and trying to touch type on a close variant, such as the azerty keyboard used in france (fig. 4B), is familiar with the effects of gesture morphing. Since some letters have been moved away from their "correct" positions, the gesture used to type these moved letters no longer gives the correct result. When an azerty touch typist attempts to use the qwerty keyboard, they experience the same effect. The deformation of azerty with respect to qwerty is both an appearance deformation and a posture deformation. On an ambiguous keyboard, the appearance can be deformed without deforming the posture. For example, on a standard telephone keypad, the letters A, B and C are assigned to key 2. Typing any of these letters requires the same gesture: and 2, pressing keys by stretching a hand. If the key is shown in reverse alphabetical order as a CBA, the appearance will be changed, but the pose will not be changed.

The optimization regarding gestures must not only take into account the appearance of the keyboard, but also the way the user interacts with the keyboard. This way of interaction will be referred to as the interaction mechanism. A diagram of an exemplary pose deformation constraint is shown in fig. 15.

How much postural deformation is acceptable?

azerty is initially somewhat difficult for a qwerty typist to touch type, but initially feels sufficiently similar to the qwerty used by the qwerty typist. By contrast, everyone immediately realized: without training, the qwerty typist cannot touch type Dvorak. This indicates that there is some non-zero threshold of appearance distortion that is acceptable for inexperienced consumers without losing interest. An object of an aspect of the present invention is to achieve improvement in typeability with such a small margin. It is not overly emphasized that most commercial failures of prior art innovations are due to the lack of recognition of the deformation limit, let alone compliance with the deformation limit.

In the azerty-qwerty variant, 5 letters are shifted. All of these are changes involving equal or near-equal finger movement. Four of the letter movements are represented as two exchanges. The experience obtained may be: if the keyboard is to be used immediately without training (with possible typographical errors), then 5 important gesture changes are an upper bound on the allowed gesture deformations. Psychological studies need to be performed to give a better boundary than this one obtained from the thinking according to the prior art.

For symbolic representation of deformation

Recall that the problem to be solved by the present invention is to minimize the negative impact of distortion on consumer acceptance of new keyboard products. The basic implementation is that if the deformations can simply be symbolized, the deformations can be made to resemble better and thus be minimized. The simple symbolic representation makes it easy to interpret, memorize, compensate for deformations.

A well-known method in computer science to measure the complexity of an object is to calculate the length of the shortest order required for the object. The deformation can be measured in the same way. The description is a set of words sufficient for a person who knows the words to find each and every letter on the keyboard in conjunction with any conventional knowledge known to those skilled in the art. Imagine a salesperson explaining to a potential customer a new keyboard, for example, "it is similar to qwerty but a and z are reversed" might describe a first keyboard, while "it is similar to qwerty but a moves two keys to the right, r moves two keys down, t moves two keys to the left and one key down" might describe a second keyboard. In this case, the first keyboard is less distorted than the second keyboard because the description of the first keyboard is shorter.

Associated with the description length are other ways of symbolizing the deformation. Memory (menemonics) may be useful because it may be the association of morphs with known words, sounds, or objects. Indeed, any known memory method may be useful in expressing the deformation in a manner that makes the deformation more desirable to the consumer's taste. Several possible symbolic representations of variations and their use in designing keyboards will be discussed in the detailed description of embodiments of the present invention below.

Method for generating typeable optimized keyboard with minimized distortion

Referring to FIG. 16, a method of generating a typeability optimized keyboard with minimized distortion is described.

Step 1600: selecting a conventional keyboard layout

Step 1601: selecting reduced spatial arrangements

Step 1602: selecting a measure of deformation

Step 1603: selecting typewriteable metrics

Step 1604: evaluating the (typeability, deformation) measure against a set of layouts

Step 1605: selecting a layout that optimizes typeability while respecting distortion constraints

In one set of embodiments below, the method will be performed in a variety of environments, under a variety of design constraints, to illustrate its broad applicability.

Best mode for carrying out the invention

Fig. 17 presents a diagram giving an overview of the embodiments presented in detail below. The various embodiments were chosen to emphasize one or more aspects of the invention, and thus delineate the scope thereof. After absorbing the teachings of these embodiments, it will be apparent to those skilled in the art how to construct intermediate and hybrid cases that do not depart from the spirit of the disclosure, although these cases do not depart from the spirit of the disclosure.

Actual Qwerty-like keyboard for cellular phone

This embodiment is an exemplary example of how the teachings of this embodiment can be applied under realistic engineering conditions, where several constraints can be valid at the same time. It will show how various tradeoffs between typeability and morphing are managed to make it compliant with industry specifications.

What is desired here is a telephone that maximizes typeability and minimizes distortion. The deformation in appearance is suitably measured in the following manner:

1) only the number keys (0-9) of a standard telephone keypad are used for letters.

2) The read order of qwerty must be preserved as far as possible starting from the left. Specifically, the name "qwerty" must be at the beginning of the top row, with all letters in order.

3) No more than 4 letters on any key. This constraint is due to practical limitations on the number of letters that can be included in a key label, which allows for the small size of the keys and the belief that such partitioning limitations will reduce the apparent distortion.

4) The description of the keyboard in the user manual in english must be as short as possible and easy to remember. This constraint is adopted because of the cost of producing the user manual, and it is believed that it will reduce the effective appearance distortion.

Referring to fig. 18, we see that one way to find a solution to these requirements is to:

step 1801: typeability is maximized using only the line-preserving transform and the order-preserving transform.

Step 1802: a subset of layouts is selected that a) have the best typability and b) have no more than 4 letters on one key.

Step 1803: the respective layout from step 1802 is deformed in all possible ways by moving the 1, 2.. n letters from their original positions, placing them on the right side or 0 key of the keyboard. To maintain the original reading order, the letters are not shifted to or from the left column of the keyboard or any of the letters q, w, e, r, t, y are shifted.

Step 1804: from the layouts of step 1803, a layout is selected having a) high typeability, b) short, easy to remember specifications.

It should be understood that the problem may be solved in other ways, such as by using stochastic optimization techniques (e.g., simulated annealing algorithms or genetic algorithms). This process has the teaching advantage of bringing about the interplay of morphing optimization and typeability optimization, and is easy to perform in practice.

Step 1801: typeability is maximized using only the line-preserving transform and the order-preserving transform. This can be accomplished, for example, using any of the methods described in Gutowitz' 317. Here, our primary goal is to study the relationship between layout scope and typeability. For equal typeability, a lower layout range is preferred. To accomplish this, we will optimize the typeability (here, measured in terms of the number of significant keys) of each layout in a set of layouts where the layout range is fixed at 1 to 7.

The results of applying this step are shown in fig. 19 and 20. In fig. 19, the number of valid keys of the optimal layout found for each min-max range from 2 to 7 is represented as a function of the range. To guide the interpretation of these results, several horizontal lines are drawn. Reading from bottom to top, these lines give:

a) the number of valid keys for the qwerty class code that is coded as uniformly as possible over three columns. The layout of the coding as uniform as possible is shown in fig. 14.

b) The number of significant keys of a Standard Ambiguous Code (SAC), i.e., the "abc" code of a conventional telephone keypad.

c) The minimum number of valid keys for type A touch typability defined by Gutowitz' 317.

d) The number of valid keys on the 9 keys that allow the letter to be arbitrarily assigned to the best possible code for the key.

e) As in d), but for the 10-key code.

The layout corresponding to the dots drawn in fig. 19 is shown in fig. 20, in which the layouts in the range of 2 to 7 are shown in fig. 20A to 20E, respectively.

We note that these results show that there is no advantage in considering the range of 4 or more in terms of typeability. Increasing the range not only increases the distortion, but also appears to decrease the typeability. To further address this issue, we may then be limited to studying layouts with a range of 4 or less.

Note that the curves of the optimal layout never pass the a-level touch typable line. Thus, the experiment shows that touch typable codes on a telephone keyboard cannot be obtained if the row and order constraints are fully respected. The division distortion alone is still sufficient to substantially improve the typeability above the base level set by the most uniform coding possible.

Step 1802: a subset of layouts is selected that a) have the best typability and b) have no more than 4 letters on one key.

The need to meet this requirement stems from the observations just made: a large range may reduce the typeability. In this case, the explicit deformation restriction and the restriction on improving the typeability are coherent with each other. It will be appreciated that this is not generally the case: increasing the allowable distortion increases the level of achievable typeability.

Step 1803: the respective layout from step 1802 is deformed in all possible ways by moving the 1, 2.. n letters from their original positions, placing them on the right side or 0 key of the keyboard. No letters are moved from the left column of the keyboard or any of the letters q, w, e, r, t, y are moved.

After the partitioning transformations are performed as much as possible, step 1803 explores the effect of adding a small number of order transformations. Sequence distortion is limited by the desire to minimize the perceived distortion.

The result of this step is shown in fig. 21. Here, the distribution of the number of effective keys of the layout generated with 1 to 4 order transformations is shown. It can be seen that as the number of order variations increases, the distribution of the number of effective bonds becomes wider. Although the average number of valid keys remains approximately equal as the number of order variations increases, it becomes possible to find layouts with increasingly better (worse) numbers of valid keys at the extremes of the distribution.

In step 1803, the letter is allowed to move to the 0 key, violating both the row and order constraints, and potentially increasing the number of letter keys to 10. It also allows all letters on a key to be moved to other keys, which reduces the total number of letter keys. Thus, fig. 22 shows three curves, one for each of the 8, 9 and 10 letter keys. The number of valid keys for a given number of order variations and an optimal layout of a given number of keys is shown in these curves. These horizontal lines are the same as those of fig. 19, but with the addition of a line giving as uniformly coded as possible a significant number of keys in 5 columns. The most uniform coding possible over the 5 columns is shown in fig. 14.

As can be seen from fig. 22, with the sequence morphing, a-level touch typability of the phone keys can be obtained with 9 or 10 letter keys (although not 8 letter keys). In fact, with 10 letter keys, a touch typability level substantially the same as 5 columns of as uniform layout as possible can be achieved, with as few as three sequence variations. But are three sequential distortions acceptable low level appearance distortions? How to attenuate the visual impact of these sequential deformations? This will be addressed in the next step of the process.

At step 1804, a layout is selected from the layouts of step 1803, the layout having:

high typeability, and

the short description length is the length of the short description,

easy to remember description.

To balance this trade-off, we first set out the constraint of a short description length. To quantify this constraint, we will consider the following form of layout specification: "it has a qwerty layout except: [ itemization of exceptions ] ".

Any variant qwerty keyboard can be described in this format. The length of the description is related to the following factors: a) the number of exceptions, and b) may express the conciseness of the exceptions. A typical exception will be written as: "except that the letter x is on the 0 key".

It is noted that if two letters are moved to the same key, two exceptions, e.g. "except the letter xy on the 0 key", may be represented without doubling the number of words.

Such rules would be easier to remember if the letters were not arbitrary, but rather could be pronounced, or could better spell a word such as "lu" or "gum", e.g., "except 'gum' on the 9 key". This has the same contents as the following item "except that the letter g is on the 0 key, the letter u is on the 0 key, and the letter m is on the 0 key", but is easier to remember.

The foreground candidate from these considerations is the "qwerty-GLU" layout of fig. 23, and is labeled as the point "GLU" in fig. 22.

The layout has three sequential variations. The letters g, l and u are not in their qwerty positions. They are moved to the end of the layout. Thus, the main part of the layout is read without insertion of only deletions, and the deleted letters reappear at the end of the reading order. The letters "GLU" are pronounceable, appearing in order of their skill-type pronunciation, and form part of a memorable mnemonic, "qwerty" with GLU (sticky) to cell phone. The number of valid bonds is very close to the maximum number obtained in this experiment for any layout with three order variations.

It will be appreciated by those skilled in the art that the process allows many variations, while still remaining within the scope of the present invention. Different constraints may be used. The steps may be performed in a different order or some steps may be omitted. Different basic conventions other than qwerty may be used. Different keyboard geometries may be used, as well as different memoiy techniques.

Applying the method to 5 columns of qwerty

It will be appreciated that the above-explained method for finding a qwerty-like keyboard with optimized typeability and minimized distortion for a telephone keyboard may be modified to apply to many situations. In this section, we will quickly examine the results of applying this method to construct the layout of a 5-column qwerty-like keyboard. Although in the case of a telephone keypad work is required to find an acceptable keypad with a-touch typability or better typability, with a minimum order distortion in the case of 5 columns, typability of C and above can be obtained.

Turning now to FIG. 24, we see the result of applying the method of FIG. 18 to a 5-column qwerty class keyboard. This figure is substantially the same as figure 22, except that it is now applied to a 5-column qwerty class keyboard instead of a 3-column qwerty class keyboard. Because of the greater number of active keys being discussed, we are able to consider the relationship of these keyboards to the touch typability of the higher levels (i.e., levels B and C of Gutowitz' 317). When the most uniform keyboard on 5 columns has typability between class a and class B, touch typability higher than class C can be obtained with only division deformation without order deformation. As the number of sequence variations increases, the level of touch typability also increases, as can be expected from the results just presented for the 3-column keyboard.

Turning now to fig. 25, we see details of the various layouts corresponding to the points on the curves of fig. 24. For comparison, fig. 25A again shows a most uniform keyboard over 5 columns. Fig. 25B to 25E show the keyboard with the amount of sequence deformation added. For fig. 25B to 25E, the shifted letters are (none), (u), (di), (diu), (lguh), respectively. Notably, it may be felt that fig. 25B without order distortion is more distorted than the appearance of fig. 25C with one order distortion. Fig. 25B has a larger range, which is 3 because the maximum number of letters on the key is 4 and the minimum number is 1, whereas in fig. 25C, the maximum number is 3 and the minimum number is 1, so the range is 2. Thus, the psychological test may show that fig. 25C is less distorted than fig. 25B. In the case of fig. 25C, a simple mnemonic, "yoU (you) to center," is available to help memorize the deformed layout.

Easy-to-remember two-key keyboard

The easiest keyboard is perhaps one where all letters are on the same key. In a sense, it is compatible with any management, and the association of letters and keys is trivial to remember. Unfortunately, regardless of how these characteristics are defined, the typeability characteristics of a one-key keyboard are quite poor.

The next step in achieving a full keyboard is a two-key keyboard. At this step, there has been a challenging problem of designing a keyboard that is easy to remember, is compatible with conventions, and has good typeability characteristics. The present invention shows how to overcome these challenges. The two-key problem has important industrial applications. Many electronic devices that may benefit from text entry do not even have a keyboard with as many keys as there are telephone keyboards. A typical example is a digital camera comprising a navigation keyboard. Such keyboards typically have two or more arrow keys. Text entry can be made using these keys as long as there is a sufficiently accurate, sufficiently learnable method available for such a small number of keys. Text entry would be useful, for example, to annotate photographs.

We will now present several embodiments of the invention to address the two-key problem in a manner that magnifies or enhances the disclosed teachings.

FIG. 26 illustrates, without limitation, a typical navigation keyboard. Here, there are four arrow keys that are typically associated with movement: left 2601, upper 2602, right 2603 and lower 2604. The center key 2605 is typically associated with an "accept" or "advance" action.

We will consider several methods of entering text using a navigation keyboard, all of which are quite different from each other, however all are within the scope of the invention. The method comprises the following steps:

preservation of alphabetical order.

The retention of the qwerty gesture.

The use of a pure symbolic method that is conventional independently of any layout.

Row reservation of the phone keypad.

Fig. 27 shows a three key system with two letter keys and one Next key. The Next key will be used to advance letters in the character-based disambiguation system and words in the word-based disambiguation system. In this illustration, the alphabet is divided into two halves, with half of the letters on each letter key. Other options are also possible, as will be discussed below. A possible association of these three keys with the navigation keyboard of fig. 26 is to associate the letter key of fig. 27 with the two arrow keys of fig. 26 and the Next key with another letter key or "accept" key.

Fig. 28 illustrates an alternative two letter arrangement for a navigation keyboard, where the letters on the left half of the qwerty keyboard are associated with the left letter key and the letters on the right half of the qwerty keyboard are associated with the right letter key. Fig. 28A conceptually shows the layout, and fig. 28B shows the qwerty layout superimposed on two keys. The keyboard is advantageous for experienced users of reduced qwerty keyboards that utilize a two thumb interaction approach. The gesture of the thumb is almost the same, except in the navigation keyboard version, which does not require the thumb to move between keys.

A keyboard may be designed that is optimized for the length of the caption without regard to appearance or pose distortion. Consider the 2-letter key layout of fig. 29 as a non-limiting example. In this keyboard, all consonants are assigned to the left key, and all vowels are on the right key. This last sentence describes a keyboard that is sufficient to allow a person who knows the meaning of the consonants and vowels of a word to locate all the letters on the keys. Thus, the keyboard is easy to interpret and remember, which illustrates an aspect of the present invention.

We have indicated the advantage of minimizing line distortion from the point of view of apparent distortion. The letters in the morphed keyboard should be in the same row as the conventional keyboard to which the morph relates, if possible.

Turning to FIG. 30, we see a navigation keyboard in which three arrow keys are used as letter keys. The letters associated with each key are the letters in a row of a standard telephone keypad. The letters A-F2608 correspond to (ABC, DEF) on the telephone keypad, G-O2606 corresponds to (GHI, JKL, MNO) on the telephone keypad, and P-Z2607 corresponds to (PQRS, TUV, WXYZ) on the telephone keypad. The keyboard may attract those with advanced experience typing on the telephone keyboard. The gestures used for typing on the navigation keyboard thus constructed are similar to those used for typing on the telephone keyboard. Since the association of letters with rows is carefully preserved, it is easy to interpret the keypad to those familiar with telephone keypads.

One way of assessing the typability of these various two-key embodiments, which is now familiar to the reader of this disclosure, is to measure the number of key strokes, the number of valid keys, or other characteristics associated with the disambiguation mechanism per character. We will now consider applying some new techniques to this case.

Method for predicting which two-key method is better

We have discussed the length of the description of the complexity metric as used in computer science and show how it can be applied to measure the apparent deformation. Another way to conceptualize the complexity of an object in computer science is the run time of the shortest program that computes the object. This complexity measure is also relevant to keyboard design and can be used to estimate the market acceptance of the various two-letter key embodiments presented above.

The two-letter key variants of qwerty, alphabet, and vowel-consonant appear to be substantially similar in explanation complexity. One might guess based on this that they would all have roughly equal chances to succeed in the market. To predict this accurately, it is necessary to study how well the complexity measure fits the actual human purchaser's perception. It may be the case that a consonant/vowel keyboard is judged to be easier than a separate alphabet keyboard, which in turn is easier than a separate qwerty. Nevertheless, those users who are well trained for two-thumb typing on a miniature qwerty keyboard may prefer a split qwerty.

While each of these illustrations corresponds to a short program that calculates the positions of all letters, the run time of the program can be quite long. In the case of a separate alphabet keyboard, one may have to imagine reciting the alphabet, stopping at the desired letter, and checking whether they have already recited "m". This takes a certain amount of time. A person who knows the visual appearance of the qwerty keyboard may mentally scan the keyboard for letters. A person trained to play two thumb qwerty knows the position in the movement pattern of their thumb. For example, on a 26 letter key thumb operated qwerty keyboard, the motion pattern for typing the letter Q is "move the left thumb to the key with Q and press the key". To type on the novel two-key qwerty keyboard, the mode is edited as "thumb-down". Thus, a two-key qwerty keyboard is easy for a two-thumb touch typist.

The implementation mode is as follows: exemplary embodiments of gesture preservation with radical layout deformation

Embodiments of this section illustrate that even if the layout is aggressively deformed, the pose may be preserved. The keyboard is intended to be used by the driver who is driving without having them remove their hands from the steering wheel. While aiming to balance the qwerty touch typing capabilities by preserving gestures.

Turning to fig. 34, we see a steering wheel 3401 in which a keyboard 3402 is embedded or attached, preferably in a position comfortable for typing and driving, in the steering wheel 3401.

To preserve the posture, particularly to make the morphed keyboard equi-fingered to the qwerty keyboard, all letters typed with each finger on the qwerty keyboard are assigned to the same key of the morphed keyboard. Thus, all letters q, a, and z typed with the left hand pinky using the qwerty keyboard are assigned to the same key in 3402. Note that all letters r, f, v, t, g, b typed with the same finger of the left hand (but each letter from each key column on the qwerty keyboard) are assigned to different keys in 3402. This improves gesture compatibility because on both the qwerty keyboard and the keyboard 3402, the finger must be moved to the right from its home position to type in the individual letters t, g, and b. The number of keys can be further reduced by combining these keys, with an accompanying increase in posture distortion and a decrease in typeability.

If the typability metric is a valid number of keys, the typability of any of these layouts is quite poor, however, given the teachings of the present invention, it will be appreciated that the typability is improved if the strict equi-finger or equi-column pose is preserved, for example by allowing the letters to move to the adjacent finger.

Although the keyboard is discussed in the context of a steering wheel embodiment, the keyboard is useful in any device that allows only one row of keys, with a limited amount of space available for the keyboard. An example might be the edge of a pocket-sized device such as a digital camera or mp3 player. The keyboard may be used in the handlebars of a bicycle or the like.

Drum beating effect

When it is desired to repeatedly press a single key as quickly as possible, one can typically reach 7 keystrokes per second. If one letter is entered with each keystroke, the rate will correspond to approximately 75 words per minute. However, persistent typing rates of 150 words per minute have been reported with regular keyboards, with sudden bursts (bursts) of up to 212 words per minute. Typing on a regular keyboard requires time for moving a finger between keys in addition to the time required to press a key. Even ignoring the move time, these typing speeds are too fast to coincide with the repeat time on a single key. Higher speeds can be achieved because while one finger is completing a key press, the other finger is beginning another key press. If successive keystrokes are made by different fingers, the keystrokes may occur in parallel. This is the so-called drum beating effect. It is widely believed that the Qwerty keyboard has been designed: alternate hands are used to type in the usual letter pairs, e.g. th, he, qu. We will briefly check this assertion. The design is said to aim at minimizing jamming of the typewriter trace (typebar). Maximizing left-right alternation has the advantage (that may not have been expected) of optimizing typing speed for touch typists. A pair of left and right alternating keystrokes can be executed partially in parallel; movement of the second hand may be planned and performed while movement of the first hand is complete. Even for a single hand, different fingers may move more or less in parallel.

Selecting two-key layout based on drum beating effect

In the above we consider the length of the description and the time of the inner calculation as a means of predicting which two-key layout the consumer prefers. In this section, we will make preference predictions based on the drum beating effect on these same keyboards.

Consider a simple model of the drum beating effect, where the time to sequentially enter a pair of letters is 1 if they are on the same key, and 1/2 if they are on different keys. Under this model, we can easily predict the time that a skilled user will take to enter a letter using either of the two-key embodiments discussed above. The results are shown in fig. 32. In this figure, inter-keystroke times are found for each of the 26 letter order variants 3201. In various variations, the letter preceding the given letter is on the left key and the letter following the given letter in order is on the right key. The minimum time occurs at letter number 10 (j). Therefore, we have surprisingly resulted in: splitting the alphabet at j will result in a faster time than any other split. Those skilled in the art who are not instructed to use the drum beating effect to evaluate a keyboard will likely pick a letter closer to the middle of the alphabet, such as m (as shown in fig. 27). As can be seen in fig. 32, this inter-keystroke time is less than the inter-keystroke time of the two thumbs qwerty 3202. As a surprise, the lowest inter-keystroke time in all keyboards is for the consonant-vowel two-key keyboard 3203. Recall that the argumentation that novice users like to use a consonant-vowel keyboard is: a) unlike the qwerty layout, it does not require a high level of experience for two-thumb typing on a reduced qwerty keyboard; and b) unlike an alphabet keyboard, it does not require a centric scanning of alphabetical order. Thus, in this case, the acceptance criteria of the novice user and the experienced user seem to go in the same direction, favoring the deployment of the consonant-vowel keyboard. Psychological tests are required to confirm or reject the prediction.

Optimization of drum beating effect by minimizing steric hindrance

On a very small keyboard (whether ambiguous or not), a finger (not the thumb or thumb) may share a keyboard "field" with other fingers. When the size of a finger is large compared to the size of a key, the presence of one finger on a given key can hinder the ability of another finger to occupy a nearby key. This effect is called steric hindrance.

This size effect significantly complicates the analysis of the drumming effect. Referring to fig. 31, we see a series of smaller and smaller keyboards that can be typed with two thumbs. The relative sizes of the keyboard and thumb in this figure suggest the relative sizes in the case of commercial handheld devices. It can be seen that the amount by which one thumb is obstructed by the other is sensitively dependent on the keyboard size. For relatively larger keyboards (fig. 31A), when a first thumb is placed over a key, a second thumb can be moved to any other key not directly covered by the first thumb. For smaller keyboard sizes (fig. 31B), the thumb may block not only the key it is pressing, but also movement to surrounding keys. As the size becomes smaller (fig. 31C), the obstruction extends over a large portion of the keyboard.

The drum beating effect relies on the ability of one thumb to strike a key while the other thumb is moved to its striking position. For obstruction, if the target of the second thumb is in the obstruction area of the first thumb, one thumb must wait for the other thumb to shift after the keystroke is made. The obstruction may be complete or partial depending on the size and geometry of the keyboard and the key pair to be pressed during the drum strike.

The exact way in which the fingers obstruct each other for a given keyboard depends on:

the interaction mechanism is used for the user to interact with the user,

the probability distribution of the sequence of symbols,

the spatial distribution of the bonds.

The final design of the keyboard to minimize finger blockage will depend on how well these factors are known and how well they are captured with a mathematical model. The present invention teaches the use of some models for measuring obstruction.

As a non-limiting example, we can consider a simple model of this potentially quite complex case: for the right thumb, any key directly to the left, above or below the left thumb target is considered to be completely obstructed; similarly, for the left thumb, any key directly to the right, above, or below the right thumb target is considered to be obstructed. The time for blocking a letter pair is considered to be the same as the time for two letters on the same key, the time for an unblocked pair would be 1/2 of that time,

t_ave＝1/(#(i))∑_iT(l_ll_i+1) Wherein if (l)_il_i+1) If blocked, then T (l)_ll_i+1) R, otherwise T (l)_ll_l+1) R/2, wherein l_iIs the letter, # (i) is the number of letters in the string, and r is the time of double-clicking (douple tap) on a single letter. The model is available from MacKenzie, i.s.,&inspiration from the model of Soukoreff, R.W (2002). Which is a two-thumb text input model. Proceedings of Graphics Interface 2002, pp.117-124. Toronto: the Canadian Information processing Society.

In short, any pair of letters in which the second letter is located on the same key or on an adjacent key is considered to be effectively located on the same key. In this case, a double click time is used. If two letters are not on adjacent keys, a double-click time of 1/2 is used.

More advanced models also take into account the distance traveled by the finger, in terms of the Fitts effect, partial obstruction effect, and other more subtle effects.

Optimization of drum beating by proliferation (multiplication) of common symbols

It should be appreciated that the drum beating effect in the presence of steric hindrance can be optimized by both partitioning and order deformation in the above-described manner and using a model such as that given above. Optimization can also be performed by modifying the physical structure of the keyboard. For example, the keys may be spread or otherwise shaped to increase the likelihood of entering ordered symbol pairs with a drum strike. We will now briefly discuss an embodiment that seeks to optimize the drumming effect by multiplying the representation of selected symbols, particularly when the steric effect is important. The symbol may be a common letter (e.g., letter e in english), or a common punctuation mark (e.g., space mark), or a frequently used function symbol (e.g., "Next" or "Shift").

The positions of the multiplication symbols are chosen such that one or the other representation of the symbols can often be input in drum-beating order under a given interaction mechanism, thereby avoiding steric effects.

For word-based or character-based disambiguation keyboards without shift keys, one of the proliferation symbols is preferably "Next", since a Next function is often required. When the shift key is used for disambiguation as in the embodiments discussed below, the shift key may be selected as one of the multiplier symbols.

Referring to fig. 33, we see a telephone keypad 330 with 9 alphanumeric keys 3300-3309 and two Next keys 3311 and 3312. The Next bond is multiplied, i.e., represented on more than one bond.

It is expected that character-based disambiguation will be used to select a Next function for propagation. In character-based disambiguation, the Next function will be used more often than any letter or punctuation. In fig. 33, the two thumb interaction mechanism is considered to select the key on which to place the proliferation symbol. Consider the letter "q" typed in a prior art system where there is only one Next key, say, located on the x-key of fig. 33. "q" is an uncommon letter, so it is possible that the disambiguation system will place the other letters p, r, s on the key before "q", which requires pressing the Next key 3 times to enter "q". If the keys are small relative to the size of the thumb, the keystroke sequence will be:

press the pqrs key with the right thumb.

Move the right thumb to the Next key.

Press the Next key three times.

According to our model, this sequence would take 4 double-click time units plus the time it takes for the right thumb to move from the pqrs key to the Next key.

If the keyboard is large so that the left thumb can be moved to the Next key while the right thumb is on the pqrs key, the following keystroke sequence can be used:

press the pqrs key with the right thumb.

Press the Next key with the left thumb.

Press the Next key twice more with the left thumb.

The first two steps are combined into one drum hit, and since these steps involve two thumbs, the second step takes 1/2 double click time. The total time is 3 and 1/2 double click time units.

On the keyboard of fig. 33, the sequence is:

press the pqrs key with the right thumb.

Press left Next key with left thumb.

Press right Next key with right thumb.

Press left Next key with left thumb.

Even if the keyboard is small, the time is 2 and 1/2 double-click times. Thus, propagating the Next bond substantially eliminates steric hindrance with respect to the Next bond. This will improve throughput (number of symbols entered per unit time) even on large keyboards and have a more prominent effect on small keyboards.

In general, if only one symbol can be incremented given the number of keys available on the device, it should be the most frequently used symbol (functional or other). With the hybrid chord/ambiguous encoding method of Gutowitz' 317 and the embodiments below, the shift keys are generally the best candidates to be proliferated, so that the shift keys of the embodiments below can be well represented on 3311 and 3312. It should be understood that: if the number of available keys is sufficient, the second, third, …, nth most frequently used symbols may be multiplied; and the position of these proliferation symbols in the layout should be chosen to minimize steric hindrance and maximize the drumming effect.

Optimization for more than one interaction mechanism

The user population is inconsistent. On the one hand, there are users who are reluctant to risk, who want something familiar even at the expense of typeability, and on the other hand, those who appreciate typeability are willing to engage in learning new interaction mechanisms and/or layouts to obtain it. However, to achieve economies of scale, manufacturers prefer to produce a large number of individual products and wish to attract more or less everyone in the user population. One approach is to find the lowest common denominator between various user groups. Another approach (taken here) is to attract both people who are not willing to risk and people who are eager for typeability. In some other embodiments of the present invention, we have sought to produce a single keyboard with a single layout, which keyboard has both familiarity and improvements. Another solution to this problem is shown in this embodiment, where two keyboard layouts are available simultaneously, with only software changes between them, and both keyboards are optimized as good as possible for typeability, but with different interaction mechanisms.

More specifically, we consider a layout with and without gear shifting implemented on the same keyboard. Gutowitz' 317 discloses a general method to do this, he shows how to use a combination function key (or other means of combining keystrokes in a single gesture) to optimize typeability: a new layout is effectively created from an existing layout by adding another switched-out "dimension" to the layout. The same approach will be used here, with the difference that the underlying layout is the one with the least partitioning distortion from the conventional layout. Although this embodiment is well within the scope of Gutowitz' 317, it has the particular advantage of minimal partitioning distortion from conventional layouts, thereby enabling both the radical and switched-out layouts to be optimized for typeability. This creates an appeal to a wide range of users, including users who refuse to use unfamiliar gear shifting mechanisms and users who appreciate such use, in view of the greatly improved typeability that it provides.

It should be understood that the interaction mechanisms selected for combination, while remaining within the scope of the present embodiments, may vary widely. In particular, 1-finger, 2-thumb, 3-non-thumb, thumb + n-non-thumb, and 8-non-thumb interaction mechanisms may be combined in accordance with the present invention.

To embody the idea, but not by way of limitation, consider the following set of design specifications:

the layout must resemble qwerty in appearance.

The layout must be able to fit on a standard telephone keypad.

For those users who are unwilling to use the shift key or who are unable to use the shift key because they can only type with one hand, the keyboard must be typeable and must have no worse typeability than taking standard ambiguous codes based on word disambiguation.

For those users who can and intend to use the shift key, the typeability must be as high as possible.

A single layout must be used for both of the following interaction methods: a non-thumb interaction method without shifting, a two-thumb interaction method with shifting.

In order for the typeability to be no worse than standard ambiguous codes, the number of valid keys must be no less than the number of valid keys of standard ambiguous codes, i.e., 6.0. To limit the appearance distortion, we can try to use any qwerty-like layout for the phone keyboard with only a division distortion as the basic layout and make the effective key number at least 6.0. We can then consider all possible ways of shifting a letter from each key on each layout and evaluate the number of keys available for the keyboard after shifting.

For comparison, we can also consider using one of the best phone keyboard layouts with order morphing, namely the qwerty-glu layout identified above, and again consider all possible ways of selecting one letter on each key as a shift letter.

The results are shown in fig. 35. The left side is the shift layout resulting from the out-of-order deformed layout, and the right side shows the shift layout corresponding to the qwerty-glu layout. The number of keys available for the basic layout is plotted against the number of keys available for each corresponding shift layout.

There are many interesting places in the collection. With respect to the above-described embodiments, one skilled in the art may select one or the other according to further design specifications. For example, if it is desired that the shift layout typeability concerns exceed the basic layout typeability and avoid order distortion, the layout 3501 may be selected. This layout is shown more fully in fig. 36. In this full view, the shift letters on each key are shown in italics, while the non-shift letters are shown in normal font. Similarly, if it is desired that the basic layout typeability concern exceed shift layout typeability, but no order distortion is allowed, then layout 3502 (FIGS. 35 and 36) may be selected.

If order distortion is allowed, increased typeability in both the base layout and the gearshift layout can be obtained, as shown in FIG. 35. There are many shift layouts corresponding to each basic layout. To select a single shift layout from the group of shift layouts corresponding to the basic layout qwerty-glu, we can consider the economics of the above-described constraints. The global optimal layout that only considers typability is identified as 3503 in fig. 35 and 36. We see that for layout 3503, the letter of the shift is the last letter on each of key 1 and key 7, as well as the first letter on each of the other letter keys. To minimize the illustration length, a layout may be preferred: all keys have the first letter or the last letter as the shift letter. The last letter shift of all keys is the layout 3504 of fig. 35 and 36, while the first letter shift of all keys is the layout 3505 of fig. 35 and 36. Unfortunately, in this case, short description length and typeability are contradictory. Between shifting the last letter on each key and shifting the first letter on each key, one may prefer to shift the first letter because a capital letter is typically the first letter of a word (in english) and can be obtained by pressing the shift key using a standard full-size keyboard. Therefore, 3505 is preferable. However, at 3503, 3504 and 3505, 3505 has the lowest number of valid keys. Layout 3504 is in the middle in explaining familiarity and in the middle in typeability. 3503 is excellent in typeability, but requires more instructions. Comparing the shift layout of qwerty-glu with the shift layouts 3501 and 3502 corresponding to the out-of-order deformed layouts, we see that although they have order deformations, they have smaller dividing deformations (the range of qwerty-glu is smaller). Therefore, it is actually perceived that one of the shift-related layouts of qwerty-glu is less deformed than the appearance of the out-of-order deformed layout. Only psychological tests in which participants are asked to identify the layout they consider most similar to qwerty can fully solve the problem.

It should be understood that although throughout the text we refer to "shift" as a means of unambiguously identifying one letter on each letter key, any other known means may be used, such as double-clicking to get a shifted letter and single-clicking to get a non-shifted letter, long-pressing to get one letter, short-pressing to get another letter, etc.

Predictive compensation for distortion

In the learning phase, typing errors may occur due to a mixture of conventional typing gestures and novel typing gestures as the user transitions between use of the conventional keyboard and the novel morphed keyboard. The effect is to make the unambiguous keyboard ambiguous and to introduce additional ambiguity to the already ambiguous keyboard.

Disambiguation software can be used to resolve many of these ambiguities. For example, the azerty keyboard is a variation of the qwerty keyboard used for persons trained in typing on qwerty. If such a person attempts to type English on an azerty keyboard, they often type "zhat" because "what" is a common word in English, and the letters w and z are reversed in position in qwerty and azerty. Since "zhat" is not a common word in english, disambiguation software can be designed to automatically replace each occurrence of "zhat" with "what". Although this basic idea is simple, practical difficulties arise in many cases. The user would like to type "zhat", perhaps an abbreviation. In this case, replacing "zhat" with "what" would be the wrong. Disambiguation software can have difficulty determining whether "zoo" was typed correctly or was intended to be "wood", both of which are commonly used.

The same considerations apply to character-based disambiguation. For example, the alphabetical pattern "zz" is much more frequent in English than the pattern "ww", however, replacing www with zzz in a URL would be erroneous.

Similar to training wheels (training wheels), disambiguation software can help start learning, but later become a hindrance. Therefore, it is desirable that the strength of distortion compensation disambiguation be adjustable. This can be achieved in various ways. Preferably, given the statistics of the language, a sequence of likelihoods is computed for both the regular keyboard and the warped keyboard. This calculation will be apparent to those skilled in the art of statistics and probability theory. The user adjustable parameter then sets a threshold such that sequences having a closer likelihood than the threshold are not automatically overwritten, and replaces the distorted sequence with the normal sequence when the likelihoods of the sequences are far apart and the normal sequence is most likely.

With reference to fig. 37, we describe in more detail how this aspect of the invention performs.

Step 3701: a likelihood threshold is set. The settings may be under the control of the user, or may be set in hardware or software (perhaps based on an analysis of user behavior). The likelihood threshold determines the relative weight given to a conventional keyboard interpretation or a warped keyboard interpretation of a keystroke sequence.

Step 3702: the user enters the letter sequence K.

Step 3703: the software calculates the sequence that may be desired, assuming a warped keyboard and a non-warped keyboard.

Step 3704: if the sequence is significantly more likely when interpreted as typing on a non-morphed keyboard, then the non-morphed interpretation is output, otherwise the morphed keyboard interpretation is output.

Selecting for a reduced number of shift letters or for a shift letter with a reduced probability

This embodiment provides an example of how the teachings of the present invention can now be incorporated into the teachings of Gutowitz' 317 regarding hybrid chord/polysemous coding. More specifically, the order warping and the division warping may be combined with an optimal selection of symbols to be selected using the chord mechanism. It should be appreciated that "chord" in this context may refer to any mechanism for distinguishing a subset of letters from a set of letters (e.g., a set of letters assigned to a given key).

For concreteness, but not intended to be limiting in any way, the present embodiment is described with respect to a gaming device. On this game device, the keys assigned letters are not labeled with letters at all. The main purpose of the machine is to play a game rather than to enter text, and the keys are labeled for gaming purposes. Therefore, as with the other embodiments, it is important for this embodiment that the assignment of the letters to the keys be easy to learn and remember.

This serves our purpose and thus limits the number of letters that can be produced using chords. To meet this limitation, and also to optimize typewritability simultaneously, order morphing and division morphing must be carefully selected.

Turning to fig. 38, we see a keyboard with a screen (3810), shift keys (3805), a set of directional keys (3806 and 3809) designed to be operated with the left thumb and a set of keys (3801 and 3804) designed to be operated with the right thumb. If we use a conventional order for the english alphabetic order and a conventional division for the standard division of letters on the telephone keypad, we can map this convention onto the gaming device in the following way: let each of the four keys (3801-3804) represent one key of the telephone keypad and represent the other four keys of the telephone keypad when they are activated together with the shift key (3805). For example, (abc, def, ghi, jkl) may be assigned to (3801-. This code has exactly the same typeability as the standard ambiguous code on the telephone keypad (6.0 valid keys, using our standard statistics for english). In fig. 38A and 38B, the assignment is shown on the screen (3810) as an aid to the user. Preferably, the display may be turned off if the user becomes sufficiently skilled that it is not necessary to be reminded of the assignment of letters to keys. It should be understood that in the present embodiment, other symbols instead of or in addition to the letters a-z may be assigned to the keys. It should also be clear that mechanisms other than shift keys may be used to distinguish the subsets of symbols assigned to the keys, and that conventional orders other than alphabetical order may be used as the basis for the present embodiment.

We can find alternative assignments that a) improve the typability as measured by the number of valid keys, and b) improve the learnability as measured by the number of letters that need to be remembered, which are associated with the shift key by:

a) all possible partitions that produce alphabetical letters such that there are 8 non-null partition elements,

b) selecting a partition having:

-i) a high number of effective bonds,

ii) as few letters as possible in the shift pattern,

iii) to the extent possible, the number of letters in the dividing element alternates between being greater than the mean and being less than the mean. This helps to achieve ii) while reducing sequence distortion.

Applying these criteria allows us to find letter-key assignments that are optimized for both typeability and learnability. An example layout is shown in fig. 39, where unshifted letter-key assignments, shifted letter-key assignments are shown on the screen (3810) in fig. 39A and 39B, respectively. The code is abcd-efg-hijkl-mn-opqr-s-t-uvwxyz in alphabetical order. Based on our reference statistics, its number of valid keys is 6.8 and the lookup error rate is 42. This is a significant improvement over standard telephone keypad coding. When the elements are alphabetical, the dividing elements above the average and the dividing elements below the average almost alternate and can be made to alternate in minimum order distortion: in the non-shift mode (fig. 39A), assigning (abcd, hijkl, opqr, uvwxyz) to the key (3801-; while in the shift mode (fig. 39B), (efg, mn, s, t) is assigned to the keys (3801-. Thus, there are only 7 shift letters, reducing the amount of memory required to learn the code relative to a standard telephone code. It should be understood that the limitation of alternating of a dividing element that is larger than the average and a dividing element that is smaller than the average and the reduction of the number of letters in the shift pattern are beneficial for learnability, but may conflict with the optimization of typeability. Turning now to fig. 40, we see a table of codes that are adapted to this situation, but where the number of shift letters varies from 4 to 12. For each number of shift letters, the ambiguous code with the highest number of significant keys is shown. The total probability of shift letters and the alphabetical set of shift letters are also shown.

Alternative embodiment based on minimizing the probability of a shift letter

The code of fig. 39 corresponds to a row with 7 shift letters in the table of fig. 40. Its number of significant keys is the highest in the sample, which is why it was selected above. If the learnability constraint is determined to be more important than the typeability constraint, the code in FIG. 40 with four shift letters may be selected instead. The coded shift letter "erst" is particularly simple to remember. Unfortunately, the four shift letter code has only a 5.6 number of valid keys, even less than that of the standard telephone code. In another case, typeability may be judged as best measured by the smallest probability of a shift letter incorporating a high effective key number. This will result in the selection of the five shift letter code of fig. 40 with a shift letter lowest probability of 0.33 in the code of fig. 40. Its effective key number 6.1 is just larger than that of standard telephone coding. Intermediate weighting of different criteria may result in selection of yet another code. Any such options, including minimization of distortion and maximization of typeability, are within the scope of the present invention.

Variable layout embodiments

So far, we have considered a solution that minimizes the distortion and optimizes the typability for a single keyboard. However, a given person may have several devices with different keyboards for which it would be advantageous to have a layout that minimizes the differences between the keyboards. Therefore, it would be advantageous to maximize typeability and minimize distortion across a range of keyboard geometries. One way of providing this is illustrated, without limitation, by the embodiments to be described now.

This embodiment is such that:

a) use the same order distortion for all keyboards in the sequence, and

b) optionally, when operating the keys in combination to select a letter, the same combination is used for the same letter for all keyboards of interest.

As a non-limiting example, consider the case of an n-column qwerty keyboard. A keyboard sequence is envisaged in which all of these keyboards are intended to be operated by the same person, potentially in rapid succession. It is desirable that they can easily and efficiently use any of the keyboards in the sequence without having to retrain the user's reflexes.

To fix one end of the range, we take the 3-column qwerty class keyboard of fig. 23 as a non-limiting example. The keyboard can be "expanded" to a non-ambiguous 10-column keyboard with the same order variations as shown in FIG. 43. Between these extremes are a range of keyboards each with the same sequence variation, each element in the range fitting different device forming factors, although potentially different divisions. For example, a device whose primary function is a telephone or the like may use a 3-column version, a primarily handheld data terminal device may use a 4-6-column version, and a device with a laptop or the like may use a 7-10-column version.

Turning now to fig. 41, we see the relative effects of order morphing and division morphing for a range of keyboards. As a non-limiting example, we will use the number of valid keys as the typeability metric for this embodiment. The first column of the table of FIG. 41 gives the number of columns for the qwerty keyboard, followed by the number of keys assigned letters in parentheses. It is particularly noted that the 3-column keyboard is listed as having 10 letter-assigned keys, which correspond to the letter-assigned keys of fig. 23. There are four columns of data in the table of fig. 41. Each data entry follows the following format: number of valid keys (lookup error rate). The number of valid keys and the lookup error rate are calculated from the same reference data as used throughout this disclosure. The first column of data, labeled EAP, gives the best qwerty class code without order distortion but with as uniform a layout as possible. The second data column labeled EAP-glu gives the value of the best order-warped keyboard (with as uniform a division as possible) with the same order-warp as the keyboard of fig. 23. The third column of data, labeled non-EAP, gives the result of finding the layout for which the best qwerty class is not as uniform and out-of-order deformed as possible. The fourth data column, labeled non-EAP-glu, gives the result of the best non-as-uniform as possible partitioning with order morphing as shown in fig. 23.

Several comments about the table are in turn:

a) these data indicate that order distortion and division distortion can be synergistically combined to produce a more typeable keyboard. For all keyboards in the range of 3 to 7 columns, either one of the order variations or the division variations alone may improve the typeability of the keyboard, but neither of the individual is effective in combination. We can easily expect that this effect will be observed for other keyboards of different layouts as well.

B) while the 3-column, as uniform as possible, out-of-order warped keyboard is less typable than the standard ambiguous coding, one of the order warped or divided warped is sufficient to produce a better qwerty-like coding under the geometry than the standard ambiguous coding, and the two together, if used, produce a touch-typable keyboard that is even strong in terms of Gutowitz' 317.

C) the keyboard with the order and division variations of 6 columns has better typeability than the hybrid chord/polysemous coding of Gutowitz' 317 when applied to a telephone keyboard, as does any variant with 7 columns. They achieve these results without using shift keys but with more keys assigned letters. Very briefly, the order morphing and division morphing used with the qwerty class layout, when applied to alphabetical order, gives results comparable to the best mix of chords and ambiguous codes. As already discussed in detail for other embodiments, all three methods of order morphing, division morphing, and chord blending may be synergistically combined together to produce further typewriteability improvements for any of these layouts.

D) it would be advantageous to use the same shift letters for all keyboards in a given variable layout family. In this way, gestures used on one keyboard in the family can be used directly on another member of the family. In some cases, a shift letter is located on the same key as a non-shift letter in one member of the family, but on its own key in the second member of the family. In this case, entering the letter in the second member of the family does not necessitate a shift. The software may be configured to cause the shift state or non-shift state to enter the same letter without causing the user to potentially quickly alternate use of any member of the family.

Turning now to fig. 42, we see how an expanded variant of the keyboard of fig. 23 is associated with devices of different form factors. Fig. 42A shows the 3-column layout of fig. 23 mated to a phone, fig. 42B shows the 5-column layout mated to a phone-like device that also has some data characteristics, fig. 42C shows the 6-column layout mated to a main data device, and fig. 42D shows the 7-column layout mated to a laptop-like device. A user who is familiar with the order transformation by using any of these devices can directly adapt to any other of these devices. At the same time, the designer of the device may select a keyboard with an acceptable key size that yields the best possible typeability for his device. This is in sharp contrast to the prior art, where designers of small devices attempt to plug a full qwerty keyboard onto the device by making the keys unavoidably small.

To emphasize this, we now turn to FIG. 44. FIG. 44A illustrates a prior art handheld data device having a full qwerty keyboard. Fig. 44B shows the same apparatus modified according to the present invention supporting a 6-column layout. FIG. 44C shows two keys of the prior art device of FIG. 44A stacked on top of a single key from the novel device of FIG. 44B. It can be seen that the novel key is much larger than the prior art keys and is therefore easier to press with an adult's non-thumb or thumb.

In view of the foregoing description, in connection with the previously discussed embodiments, it should be clear that within the general framework of this aspect of the invention (which seeks to preserve order variations across a range of keyboards), many variations are possible which remain within the scope of the invention. For example, the keyboard of the above sequence is designed to maximize typeability across all keyboards in the sequence, and the partitioning distortion is selected based only on typeability with respect to word guessing. Typewriteability may also be optimized in addition to or instead of some other disambiguation mechanism. The partitioning may also or instead be chosen for some elements of the sequence to be as uniform as possible, to have a small range, to be symmetrical, or to have some other criterion that need not be the same for all elements in the sequence. It is also clear that while the keyboard of the sequence is thought to be designed for qwerty and english, any conventional keyboard and any language group can be treated in the same set of ways as taught herein.

Implementation of variable layout embodiments

Implementation of the variable layout embodiment presents a number of ancillary problems that can be addressed by applying additional inventive insight. Three broad categories of problems and solutions thereto will now be disclosed. These problems, while particularly serious in the case of variable layouts, can also arise without involving a broader range of variable layouts. These three types of problems are: 1) the assignment of punctuation and numeric symbols to keys; 2) definition of user functions that facilitate word-based disambiguation or context-based disambiguation; and 3) the assignment of symbols from multiple languages to the same set of keys simultaneously.

Assignment of punctuation and numerical symbols

Gutowitz and Jones' 264 (incorporated herein by reference and dependent thereon) disclose a memorable scheme for assigning punctuation to keys to maximize morphological and functional similarity between symbols, particularly between punctuation symbols and numbers. A problem to be dealt with when applying the' 264 invention to the variable layout embodiment of the present invention is that the number of keys varies. Specifically, the number of keys may be greater or less than the number of digits. In the case where the number of keys is less than the number of digits, one strategy is to place several digits on one key and provide some mechanism for selecting which digit is needed. In this case, the punctuation-numerical association of' 264 can be applied directly; each number assigned to a key will be morphologically similar to a punctuation assigned to the same key. Where the number of keys is greater than the number of digits, the morphological similarities taught by' 264 may still be used to select an assignment of easy-to-remember and discoverable symbols to keys. The preferred solution for the variable layout implementation is to extend the concept of numbers to a "number pattern" and the concept of punctuation to a "punctuation pattern". The symbols in the digit mode are the preferred digits themselves or digit-like symbols in the discoverable sense. Similarly, a symbol in punctuation patterns is a punctuation symbol itself or a symbol that can be found to be "punctuation-like". By selectively adding symbols to both modes as the number of keys of the layout grows, morphological similarity between numeric symbols and punctuation can be extended to cover the entire range of variable layout sizes.

A non-limiting embodiment of the layout produced by this method is shown in fig. 45. In fig. 45, each of the keys 4501-4518 is capable of inputting symbols in any one of the following four modes: lower case letters, upper case letters, numbers, and punctuation. The keyboard is equipped with mode keys 4520, 4522, 4523 to enter the keyboard into numeric, punctuation and capital letter modes, respectively. It also has a Next key 4519 which is used to generate the Next ambiguous word or the Next ambiguous character depending on whether word-based disambiguation or character-based disambiguation is used in the current mode.

If the mode key is not pressed, the keyboard is in a default lower case alphabet mode. Each of the keys 4501-4518 includes an upper region and a lower region. The symbols in the numeric and punctuation patterns are shown in the upper region and the symbols in the alphabetic pattern are shown in the lower region. To enhance the relationship of the numeric mode symbols to the numeric mode keys 4520 and the relationship of the punctuation mode symbols to the punctuation mode keys 4522, the numeric mode symbols are located to the left of the upper region of each key and the numeric mode keys are to the left of the keyboard. Similarly, the punctuation pattern symbols are on the right of the upper region of each key, and the punctuation pattern keys are on the right of the keyboard.

There are 18 letter keys in a 6-column keyboard. In the numeric mode, once the number itself is assigned to a key, 8 keys remain. There are two assignments of additional symbols relative to the digit pattern, sum #, which follow the functional similarity approach of' 264. Both symbols are collectively referred to by the telecommunications engineer as "numbers" because they appear in a standard telephone keypad layout. Symbols (periods) are often used to punctuate numbers and so can be understood to be relatively small in functional variation from the numbers themselves and thus easily be written as part of a numerical pattern. National currency symbols are also commonly associated with numbers and thus functionally belong to the digital model. In the non-limiting embodiment of FIG. 45, the device shown is envisioned to target the United states market, and therefore, a dollar sign is shown in the numeric pattern on key 4502. On key 4502, the dollar symbol is paired with a & in punctuation mode because the & and dollar symbol are morphologically similar. The assignment of the number patterns to the remaining four keys will be discussed below in the context of a function directed to word-based disambiguation or context-based disambiguation.

In punctuation mode, the teaching of' 264 is applied directly, with 10 punctuation symbols associated with numbers. Four additional punctuation marks are associated with corresponding members of the numeric pattern on the same key to maximize morphological and/or functional similarity. Thus, pairs (#, #), (#), (#),) and ($, &) are associated with keys 4517, 4518, 4501, 4502, respectively. The punctuation pattern symbols for the remaining four keys will be discussed below in the context of a function directed to word-based disambiguation or context-based disambiguation.

For layouts in a family of variable-range keyboards with a greater number of keys, still other symbols can be added to both the numeric and punctuation patterns, following as good as possible the morphological and functional similarity scheme established by the original set of 10 (numeric, punctuation) pairs. Conversely, layouts in families with fewer keys have fewer symbols in both modes.

With this non-limiting example in mind, we can now set forth a direct teaching of the use of the' 264 invention for variable layout implementations.

To the extent possible, symbols in a numeric pattern are similar in shape and/or function to numbers.

To the extent possible, the symbols in the punctuation pattern are similar in shape and/or function to punctuation and are related in shape and/or function to one or more symbols in the numeric pattern on the same key.

For a keyboard in the family of variable layout keyboards with a key number of n > m, the set of symbols in both numeric and punctuation modes includes the set of symbols for a keyboard in the family with a key number of m.

If there are separate mode keys available for the number mode and the punctuation mode, it is preferable to place the mode key for numbers on a side of the keyboard corresponding to the side of the key on which the number symbol is placed, and to place the punctuation mode key and the punctuation symbol accordingly. With fewer available keys, several mode changing functions may be assigned to a single key.

Definition of user functions to aid word-based or context-based disambiguation

When word-based or context-based disambiguation is available alone or in combination with character-based disambiguation, it is desirable to provide a variety of functions to: a) managing changes between word-based or context-based disambiguation and character-based disambiguation; b) managing a list of words that are truly ambiguous; and c) managing the user dictionary (if any).

One aspect of the present invention is to provide these functions in a manner that:

compatible with variable layout embodiments of the invention, and also with fixed layout keyboards,

i) providing as many functions as possible, including the most important functions,

II) there is no need to perform more than one function for a single keystroke or gesture, and also provide functions to be selectively combined,

III) assigning functions to keys in a perceptible and easy-to-remember manner,

IV) is arranged to perform these functions easily using two thumbs in combination, especially in terms of steric hindrance.

To understand how these desired features may be innovatively implemented, we will now consider a set of functions to be provided as non-limiting embodiments, and non-limiting embodiments in which the functions are assigned to members of the variable layout keyboard family.

We can arrange these functions into five broad groups.

A display management function:

the next ambiguous word

The next ambiguous letter

Delete words from display

Deleting characters from display

Completion word

Prediction mode management function:

enter an alternative text entry mode

Enter basic mode (home mode)

Undo the last retrospective change

Character mode management function:

enter punctuation mode

Enter digital mode

Enter into the caps mode

Return to the base mode

Make mode viscous/non-viscous

The dictionary management function:

inserting words in a dictionary

Deleting words from the dictionary

Reordering ambiguous words

Additional management functions:

enter preference Menu

Enter additional function Menu

First consider a group of display management functions. Each of these functions operates on the current word being entered or just entered. With a word-based or context-based disambiguation system, a keystroke sequence is entered and compared to a dictionary of reference words. Several different events may occur and each requires a different action to be taken by the user. Non-limiting examples of such events and required actions include:

event: there is only one word in the dictionary that corresponds to the keystroke sequence and it is the intended word. The actions are as follows: the user only has to continue typing.

Event: there is only one word in the dictionary that corresponds to the keystroke sequence, but it is not the intended word. The actions are as follows: the word is erased and re-entered using a different input method (either a nonsense method or a character prediction mechanism).

Event: there are several words in the dictionary corresponding to the keystroke sequences, which include the intended word. The actions are as follows: the list of verbs is scrolled until the desired word appears.

Event: there are several words in the dictionary that correspond to the keystroke sequences, but none are the intended words. The actions are as follows: scrolling through the entire list of words until it is verified that the word is not found. The word is then deleted and re-entered using a different input method.

Event: the user recognizes that a typographical error has occurred. The actions are as follows: characters are deleted one by one until the result of the wrong keystroke is deleted.

Event: the user expects the system to be able to correctly complete the word based on the first few characters. The actions are as follows: the word completion function is activated.

Event: the user expects that the correct word will not be displayed even if all the keystroke systems are correctly entered, because it has performed a non-promising traceable change. The actions are as follows: the last retrospective change is undone and an alternative text entry mode is entered.

These actions include at least one display management function, but may also include other functions, such as predictive mode management functions. Three prediction mode management functions are listed above, but other functions are certainly possible. An alternative input mode needs to be entered, for example, when the intended word is not in the dictionary, so word-based disambiguation will not work and context-based disambiguation may not work. The user may also be provided with the ability to re-enter the base mode. The "undo last retrospective change" function is detailed in' 264. This functionality has the effect of helping the user avoid deleting the entire word if it is deemed that word-based or context-based will not work for correctly displaying the intended word. It only undoes the last retrospective change, leaving the previously entered beginning of the word unchanged.

The set of character pattern management functions is relatively straightforward. Assuming that all numbers, punctuation and letters are assigned to the keys as described in detail above, it is preferable to allow the user to select which of these types of symbols will be entered. It is therefore preferable to provide the user with the functionality to enter the number, punctuation and uppercase modes and to return to the basic mode, which in this embodiment is the lowercase mode. Preferably, a function is provided to "glue" any given mode to be set to the keyboard, so that the keyboard remains in that given mode until "unstuck" by another function. A common example of such a function is the Caps Lock function. However, any mode may be locked, and there may be a different function of locking each mode or a generalized function applied to each mode at present.

Word-based disambiguation systems rely on dictionaries of words. A dictionary of no limited size may contain all words (or more generally sequences of symbols that a user may wish to enter). To alleviate this problem, users may be provided with the ability to add new words to the dictionary. Therefore, a function of inserting a word in the dictionary can be provided. Conversely, it may be desirable to remove words stored in a dictionary, for example, if they are rarely used or misspelled. There may be several words in the dictionary that correspond to the same keystroke sequence. The words will be presented to the user in some default order, determined by, for example, the probability of the word, the time the word was last used, or some other automatic scheme. The user may wish to change the default order and thus may provide functionality directed thereto.

This long list of functions that help the user type with an ambiguous keyboard is still incomplete. Even with a keyboard with many keys, these additional functions may need to be available from a software generated menu rather than from the keyboard. A single keypad function would require access to additional function menus.

Further, new functions may be generated by associating basic functions into macro functions. These macro functions would be particularly useful for users who often use a given basic sequence of functions. One aspect of the present invention is to identify particular macro-functions for unusual use of word-based disambiguation mechanisms and context disambiguation mechanisms. Another aspect of the present invention is to assign basic functions to keys to maximize discoverability, usability and configurability of the keyboard.

These aspects will now be described with reference to fig. 45. The assignment of letters, numbers and punctuation marks to many of these keys is discussed above. The assignment leaves the numbers and punctuation patterns on keys 4503-4506 available. The problem to be solved is to provide as many functions as possible of the above-mentioned kind while satisfying the criteria I-IV set forth at the beginning of this section.

Consider first criterion I, which is that the layout should provide as many functions as possible, including the most important ones, directly from the basic schema. For any number of keys, there is always a trade-off between meeting criterion I and meeting the criteria of distortion minimization and typeability maximization. The keys in the basic mode may be used to provide functions or for letter assignment. The more keys used for letter assignment, the better the typeability, all other things being equal. The teachings of applying this aspect of the present embodiment must not be construed as limited to the particular keyboard of fig. 45. Indeed, we target in this embodiment: criterion I is satisfied while giving similar keyboards in the sense of belonging to the same variable layout family a similar assignment of functions to keys and of symbols to keys. It should be understood that standard I can be applied in a much broader context than the present embodiment.

For the keyboard of fig. 45, the functions Next word or Next letter 4519, enter numeric mode 4520, enter punctuation mode 4522, and enter capital letter mode 4523 are all given separate keys that make them available in the basic mode (in virtually any mode). This meets the criterion I for these functions. Other functions are available in a numeric mode or a punctuation mode, or through menus.

Let us now consider criterion II, which indicates that: there is no need to perform more than one function for a single keystroke or gesture, but rather functions to be selectively combined should be provided. To understand how the keyboard of FIG. 45 satisfies criteria II and criteria III, we will introduce eight additional functions that are available in a single gesture of pressing the numeric mode key (4520) or the punctuation mode key (4522) in combination with one of the keys assigned letters (4503-.

These eight functions are arranged in four pairs with similar functions. The first pair includes a menu entry function: enter a further function menu function, which is obtained by pressing the number mode key 4520 in combination with the key 4503; and enter a preference menu function, which is obtained by pressing a punctuation mode key 4522 in combination with the key 4503.

The second pair includes a word delete/demote function. The function of demoting words, represented by a cyclic symbol on the key 4504, is obtained by pressing the number mode key 4520 in combination with the key 4504. The function of deleting words from the dictionary, represented by the trash can on the key 4504, is obtained by pressing the punctuation mode key 4522 in combination with the key 4504.

The exact difference between these two functions may depend on implementation details and user preferences, but deleting words is clearly more aggressive than reordering words. In a typical implementation, a "delete word" will remove a word completely from the dictionary. This can be done: the deletion is limited to words previously added by the user. "demoting words" typically moves a given word to the bottom of a list of possible choices for a given keystroke sequence. For example, it may also be set to move words one bit down in the list, rather than moving completely to the bottom of the list. It should be apparent that repeated application of the word demotion function may be used to arrange the list in any desired order.

The third pair of functions changes the aggressiveness of the prediction function. The word completion function represented by the solid circle on the key 4505 is obtained by pressing a punctuation mode key 4522 in combination with the key 4505. Based on the already entered portion of the word, word completion will fill the rest of the word based on the system's best guess as to which word the user intended. This is an increase in the predictive aggressiveness. Entering the alternate text entry mode function, represented by the open circles on key 4505, reduces predictive aggressiveness. Alternative text entry modes (typically character-based predictions) are less aggressive than the default mode (typically word-based predictions). Character-based prediction only attempts to predict the next letter, not the entire word. Word completion is more aggressive than standard word-based prediction because it predicts letters even for keystrokes that have not yet been made. Entry into an alternative text input mode function is obtained by pressing the number mode key 4520 in combination with key 4505. Visual differences of full versus empty are used here to suggest more and less aggressive and to extend the theme as far as possible to other functional pairs. It should be understood that other visual differences may be used for this purpose.

The fourth pair of functions is functions that are deleted from the display. The delete word function, represented by the solid left arrow on key 4506, deletes the last word from the display, but does not remove the word from the dictionary. This function is obtained by pressing the punctuation mode key 4522 in combination with the key 4506. The delete character function, represented by the open left arrow on key 4506, deletes the last character from the display, but does not change the dictionary. This function is obtained by pressing the punctuation mode key 4520 in combination with the key 4506. As with the assignment of functions to 4504 and 4505, these assignments relative to 4506 a) place similar functions on the same key, and b) place the less aggressive of the pair of functions in numeric mode on a given key. This extends the teachings of Gutowitz and Jones' 264 by arranging functions by functional similarity and classification. This extension is combined with the extension of digital concepts to digital schema concepts and punctuation concepts to punctuation schema concepts to meet criteria III as set forth above.

Other members adapted to variable layout families

As the number of keys increases relative to the layout of fig. 45, new functionality may be added to both the numeric mode and the punctuation mode. When the number of keys is reduced relative to the layout of FIG. 45, the functions may be combined or moved to a menu. For example, the function of deleting a word from the display can always be obtained by repeatedly applying the function of deleting a character from the display, and thus the word deleted from the display can be removed or moved to a menu with fewer keys. Similarly, the function menu and the preference menu may be combined into a single menu. Authentication applications the teachings of Gutowitz and Jones' 264, exemplified above without limitation, may help a user to adapt from one member to another member in a family of variable layouts.

Selective combination of functions

In the above exemplary list of word-based disambiguation events/actions, there are several actions that comprise the basic functional sequence. For example, when there are multiple words in the dictionary that correspond to keystroke sequences, but none are the intended words, one way can be used: a) scrolling through the entire list of words until it is verified that the word is not found, b) deleting the word, c) switching to an alternative text input method, and d) re-entering the word using the alternative text input method. If this is a generic action, the user may prefer to link the actions of b) and c) so that a single keystroke or gesture performs both. These actions should not be linked by default, since: i) complex actions are difficult for novices to master, and ii) some users may prefer to keep these actions separate or combine them in different ways. For example, another user may prefer to have a longer chain of actions, including: b) deleting the word, c) switching to an alternative text input method, and e) adding the word to the lowest position of the previously entered dictionary. Additional users may prefer the latter sequence but have the added word at the beginning of the list.

This aspect of the present invention addresses these issues for all of these users by providing easily accessible atomic functions and a mechanism to complex link the atomic functions together.

Turning to FIG. 46, we see a non-limiting embodiment of a link/unlock mechanism 4600 implemented as a link/unlock menu. The link/unlink menu allows the user to set a preselected combination of atomic functions. Preferably, it also allows the user to define combinations of atomic functions. In this embodiment, the function designer 4601 appears in the link/unlock menu 4600. It has 4 components: 1) checkboxes 4602, if selected, the items are linked and moved to the top of the menu, e.g., as are linked action sequences 4606 and 4607; 2) an icon 4603 of a first function; 3) an icon 4604 of a second function; 4) a help function 4605.

The functional designer may be used in a number of ways. A first way (which we will refer to as help-driven way) is to scroll through a list of help messages 4605. Each message is a description of: the combination of the functions of the first and second functions will serve the purpose of explaining the advantages and disadvantages of each. If the user wants to perform the action, they link the functions by checking the check box 4602. A second way to design a link is to scroll the first icon 4603 and then scroll the second icon 4604. The help function will then explain the application to the selected combination. It is noted that not all combinations of first and second functions are meaningful for text input, the menu would preferably limit the selection of the second function to only those second functions that are reasonable in view of the current selection of the first function.

Once the two functions are linked, they appear in the link/unlink menu with check boxes checked. Some embodiments 4606 and 4607 are shown. Preferably, if any one of these function combinations is unlocked by unchecking the corresponding box, it disappears from the menu, making the number of items in the menu small.

Non-limiting examples of combinations of functions that some users may prefer include:

the next character + enters an alternative text entry mode. When pressing the "next character" in a word-based disambiguation mode, it is typically the case that the user has lost confidence that the system correctly found the intended word. Thus, they may prefer to have the system enter an alternative text entry mode to enter the rest of the word. The system may be arranged to revert to word-based disambiguation when non-alphabetic characters are entered.

Enter alternative text entry mode + reply to last retrospective change. If context-based disambiguation has made a retrospective change that the user does not believe would result in the correct entry of the intended word, the user may wish to undo the last retrospective change and enter an alternative text entry mode to complete the word with more complete control.

Delete word from display + enter alternative input mode. When an entire word is deleted from the display in the context-based disambiguation mode, this is often the case when the system cannot correctly guess the intended word. The user then wants to delete the word from the display and enter an alternative text entry mode so that the desired word can be entered correctly.

Blank + enter base mode. When the basic text input mode is selected to be context based rather than character based, there may be situations as described above where a temporary fallback to a character based input mode is required. Preferably, each time a symbol such as a space is entered, a return is made to the underlying context-based mode, thereby ending the word.

Enter an alternative text entry mode + insert a word into the word stock when it is complete. When context-based disambiguation fails to display the desired word and/or the user expects that context-based disambiguation will fail for the next desired word base, they may wish to enter an alternative text entry mode and enter such entered word into a dictionary for possible future use.

Delete word from display + delete word from word stock. When context-based disambiguation apparently fails, the user may decide to use only the delete word function. In this case, the user may wish to ensure that the displayed letter sequence does not appear as a future prediction.

These and many other combinations can be made based on the atomic functions described above. A subset of such combinations may be preloaded as styles (style). That is, some sets of linked functionality may be suitable for beginners, while other sets are suitable for skilled persons, and these sets may be selectable by users without requiring them to manually link all of the appropriate functionality. Obviously, once two atomic functions are linked, they can be further linked to form a longer sequence of actions.

Optimization of two-thumb typing

Finally we consider standard IV, which indicates: it is desirable that the arrangement be such that the use of two thumbs in combination is easy to perform these functions, particularly in terms of steric hindrance. Reduction of steric hindrance requirements: any gesture to be performed with two thumbs pressing two keys (substantially simultaneously or in rapid succession) should be performed on keys that are as far apart from each other as possible.

It should be noted that the prior art has focused on making a keypad that is fast to use with a single non-thumb, or stylus. Thus, the prior art has focused on placing symbols often used together in sequence close to each other to reduce the time to move from one key to another. The exact opposite of the present teaching is that the frequently used keys in combination should be as far apart from each other as possible on the keyboard. Since one element in the sequence will be pressed with one thumb and the other elements in the sequence will be pressed with the other thumb, it is important to place the frequently used keys in combination where the thumbs do not interfere with each other. In this example, it is expected that function keys will be used more frequently than numbers. Specifically, the data indicates that the backspace key is used very frequently while actually typing. Thus, the function keys and common punctuation such as periods or commas should be placed on the top row of the keyboard, as far as possible from the mode change keys on the bottom row, if possible. Such an arrangement is shown in fig. 45. It should be understood that the arrangement of FIG. 45 may not be compatible with all members of the variable layout family. Specifically, for the 3-column layout described above, it may be preferable to arrange the numbers in a familiar telephone keypad fashion.

It will be understood that many variations may be made to the illustrated embodiments without departing from the scope of the invention. In particular, advances in natural language, conventional reference layouts, keyboard geometries, distortion metrics, hindrance metrics, drum beating metrics, or interaction mechanisms will be fully apparent to those skilled in the art in light of this disclosure.

It will be readily apparent to even those less than average skill in the art that any of the above embodiments may be used in conjunction with disambiguation patterns (flueish) added to base words or character-based, such as a) word completion, b) phrase completion, c) user dictionaries, d) cross-word prediction, e) additional keys for entering additional symbols (such as punctuation marks, shortcuts), and indeed, any disambiguation mechanism may be improved by the continued application of the discoveries and techniques disclosed in this disclosure.

The scope of the invention should, therefore, be determined not with reference to the only superset of all possible combinations of aspects of the embodiments, but instead should be determined with reference to the appended claims.

Cross-referencing

This application claims priority from PCT application No. PCT/US2005/003093 at 27/1/2005. The present application incorporates by reference and relies on the following documents: PCT/US01/30264, "Method and apparatus for the acquired entry of symbols ona reduced keypad", belonging to Gutowitz and Jones at priority date 9/27 of 2001, US 6885317, granted to Gutowitz at priority date 12/8 of 1998, US 6885317, US 6219731, US patent applications 09/856,863, 10/415,031 and 10/605,157, and all other applications entitled to their priority date.

Claims (33)

1. An apparatus, the apparatus comprising: an ambiguous keyboard comprising keys; symbols characterized by being assigned to said keys, said symbols further characterized by being divided into conceptually disjoint subsets, such that all of said symbols entered ambiguously using said ambiguous keyboard are located in one of said disjoint subsets, and such that all members of a given said disjoint subset are assigned to a given said key.

2. The apparatus of claim 1, further characterized in that the symbols are letters and the separate subsets are comprised of sets of vowels and consonants in a language.

3. An apparatus comprising an ambiguous keyboard that inputs symbols characterized by comprising at least one symbol selected from the group consisting of a functional symbol and an alphabetic symbol, and wherein said at least one symbol possesses a plurality of representations on said ambiguous keyboard, said plurality of representations characterized in that each representation is assigned to a different one of said keys, and said plurality of representations are further characterized in that said different ones of said keys are arranged to minimize steric hindrance.

4. The device of claim 3, further characterized in that the plurality of representations have a shift function, the shift function characterized by entering a selected symbol assigned to a key of the ambiguous keyboard when the shift function is activated substantially simultaneously with the key.

5. The apparatus of claim 3, further characterized in that said plurality of representations have a "next" function and said symbols are arranged in an order, said "next" function being characterized in that it is used to advance said symbols in said order.

6. An apparatus comprising a single row of keys, an ambiguous code characterized by having minimized pose distortion, wherein the pose distortion is measured relative to a conventional layout.

7. The apparatus of claim 6, wherein the gesture deformations are evaluated for a set of gestures, the gestures extracted from the set of gestures comprising an interaction mechanism selected from a group comprising a two-thumb interaction mechanism and an eight-non-thumb interaction mechanism.

8. An apparatus comprising a keyboard having one to nine columns, an ambiguous code characterized by having minimized appearance distortion for both order distortion and partition distortion, said order distortion and said partition distortion evaluated with respect to conventional layouts, said ambiguous code further characterized by maximized typeability.

9. The apparatus of claim 8, further comprising a second ambiguous code, said second ambiguous code characterized in that it is a hybrid chord-ambiguous code.

10. The apparatus of claim 8, further characterized in that the apparent deformation has a minimized description length.

11. A method for producing a reduced distortion typeability optimized keyboard, the method comprising the steps of: selecting a conventional keyboard layout; selecting a reduced spatial arrangement; selecting a deformation measure; selecting a typeability metric; evaluating a set of layouts by measuring the distortion metrics and the typeability metrics for elements of the set of layouts; an optimized layout subset is selected from the set of layouts.

12. An apparatus, the apparatus comprising: a ambiguous keyboard; keys for inputting symbols, the keys being arranged in a substantially linear array; disambiguation software, the ambiguous keyboard characterized by gesture distortion minimized.

13. An apparatus comprising an ambiguous keyboard, a symbol, keys for inputting said symbol, disambiguation software, said ambiguous keyboard characterized in that at least one typeability constraint is optimized and layout distortions are minimized, said layout distortions being measured relative to a conventional layout.

14. The apparatus of claim 13, wherein said symbols comprise numeric symbols and punctuation symbols, said apparatus further comprising an assignment of said numeric symbols and said punctuation symbols with respect to at least one of said keys, said assignment characterized in that at least one of each of said numeric symbols and said punctuation symbols is commonly assigned to at least one of said keys, said assignment further characterized in that at least one of morphological and functional similarity between said commonly assigned numeric symbols and said punctuation symbols is optimized across a plurality of said keys.

15. The apparatus of claim 14, further comprising a functional symbol, the functional symbol characterized by performing a function of: modifying at least one of an output and a functional characteristic of the disambiguation software.

16. The apparatus of claim 15, further comprising a linking mechanism to link the functional symbols into a sequence.

17. The apparatus of claim 15, further characterized by pairing at least two of said function symbols based on at least one of said functional similarity and said morphological similarity, further characterized by said paired function symbols being commonly assigned to one of said keys.

18. The apparatus of claim 13, further comprising a selection mechanism for selecting a subset of said symbols assigned to a given said key, said selection mechanism.

19. The apparatus of claim 18, further characterized in that the selection by the selection mechanism is optimized for at least one of: maximizing the typeability constraint and minimizing the layout distortion.

20. The apparatus of claim 13, wherein the typeability constraint is selected from the group consisting of: lookup errors, query errors, number of valid keys, word level ambiguity, word completion, phrase completion, drum beating probability, steric hindrance, throughput, robustness to typographical errors, number of secondary selection symbols, probability of secondary selection symbols, language generality, Fitts' rule, and number of key strokes per character.

21. The apparatus of claim 20, wherein the drum beating probability is optimized with respect to a two-finger interaction mechanism.

22. The apparatus of claim 13, further characterized in that the typeability constraint is optimized with respect to at least one interaction mechanism.

23. The apparatus of claim 22, further characterized in that the interaction mechanism is selected from the group consisting of: one non-thumb, one thumb, two non-thumbs, two thumbs, one non-thumb and one thumb, three non-thumbs, and eight non-thumbs and two thumbs.

24. The device of claim 13, further characterized by compatibility with a telephone keypad.

25. The apparatus of claim 13, further characterized in that the ambiguous keyboard comprises three rows and 1 to 9 columns.

26. The apparatus of claim 13, wherein the layout distortion is selected from the group consisting of an appearance distortion and a pose distortion.

27. The apparatus of claim 26, wherein the gesture deformation is quantified according to a gesture deformation characteristic selected from the group consisting of a same hand, a same finger, a same non-thumb, and a same thumb, adjacent fingers, and a same gesture category.

28. The apparatus of claim 26, wherein the appearance distortion is a function of at least one layout characteristic selected from the group consisting of an order structure and a partition structure.

29. The apparatus of claim 28, wherein the distortion with respect to the order is the same in a family of variable layout keyboards.

30. The apparatus of claim 28, wherein the deformation with respect to the order is quantified according to a deformation characteristic selected from the group consisting of: read order, row limited read order, column limited read order, outline, row limited, column limited, quantity, and exchange quantity.

31. The apparatus of claim 28, wherein the partition structure is quantized according to partition characteristics selected from the group consisting of: as uniform as possible, the maximum number of letters on a key, the minimum number of letters on a key, the range, the dominant class size, the left-right symmetry, the up-down symmetry, and the monotonicity.

32. The apparatus of claim 26 wherein the apparent deformation is measured as a deformation relative to the conventional layout, and the conventional layout is selected from the group consisting of a telephone keypad, a qwerty national variant, and an unicode writing convention.

33. A nonsense keyboard which has undergone a sequence morphing relative to a conventional keyboard, said sequence morphing being characterized in that it forms the basis of a family of typeability optimized keyboards.