JP3379633B2 - Active message generation method and apparatus, and recording medium recording active message generation program - Google Patents

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to electronic mail and TE.
In a communication message such as LEX, message contents are analyzed, a canonical expression is converted to an canonical expression, and an active message is extracted that extracts only active messages that have meaning to a recipient who receives the message. Method and apparatus.

[0002]

2. Description of the Related Art Conventionally, in a device for summarizing or abstracting a sentence, the sentence is analyzed, the desired summary is syntactically analyzed, and the information such as the connection relation of each sentence constituting the sentence, the conjunction, etc. The relations between sentences and the weight of sentences were done by, and the summary (or abstract) was automatically created by displaying the sentences with the highest weight in order and excluding the sentences with the light weight from the sentences.

[0003]

In the above-mentioned conventional apparatus, the connection relation of sentences cannot be determined only by the connectives and the like, and proper automatic summarization cannot be performed.

An object of the present invention is to provide an active message generating method and apparatus for appropriately displaying the message contents in a form having a minimum amount of information with little redundancy for a message receiver and a recording medium recording the active message generating program. To provide.

[0005]

SUMMARY OF THE INVENTION The method of generating an active message of the present invention involves receiving a message among messages.
Active messages that make sense for the recipients to take
A process in an apparatus for generating only, the device
Is a semantic network consisting of semantics and roles of semantics
Multiple standard messages composed of are stored,
A step for inputting a basic message, a step of morphological analysis the input base message, or the result of morphological analysis
Dependency analysis is performed, and the basic message is regarded as a semantic element.
A stage consisting of a semantic network consisting of the roles of semantics
The floor and the semantic network of the basic message are standard
When the message's semantic network is satisfied,
Determined that the basic message matches the standard message
If they match, the degree of matching of the semantic network
The calculated, and outputting the highest likelihood fixed message, and converting on the basis of the output result, the fixed message parts that are similar to the fixed message in the base message, the basic message Extracting only active content that is meaningful to the recipient receiving the message, and outputting only the active message corresponding to the active content in the converted basic message. And stages.

Further, the active message generating apparatus of the present invention has a basic message inputting means for inputting a basic message and all characters of the input basic message.
On the other hand, morpheme candidates are extracted using the morpheme analysis dictionary,
Used for each morphology after all morpheme candidates have been extracted
Node cost of each morpheme indicating degree and each morpheme
Obtain the link cost, which is a measure of the possibility of interpolation connection.
Therefore, the node cost and
And the link cost from the lowest total
Output few candidates as morphological analysis results,
Dependency analysis is performed from the analysis result and the basic message
A semantic network composed of semantics and roles of semantics
The basic message analysis means and standard messages
It is composed of Ajinomoto Network, and has the meaning of multiple standard messages.
A standard message storage means for storing Ajinomoto Networks
And the semantic network of the basic message is
When the message network of Sage is satisfied,
Judge that the basic message matches the standard message
If they match, the degree of matching of the semantic network
And a message comparing means for outputting the most likely fixed message, and a message for converting a portion of the basic message similar to the fixed message into a fixed message based on the result of the message comparing means. Stylizing means, active message extracting means for extracting only active content that has meaning to a recipient who receives the message from the basic message,
And an active message output means for outputting only an active message corresponding to the active content in the converted basic message.

The invention is also computer readable.
The recording medium includes a process of inputting a basic message, a process of morphological analysis of the input basic message, and a dependency analysis based on the morphological analysis result, and the basic message has a meaning and a meaning consisting of roles of semantics. A process of configuring a basic network, a process of reading a plurality of standard messages configured of a semantic network, and the basic message is a standard message when the semantic network of the basic message satisfies the semantic network of the standard message. It is determined that the message matches, and if it matches,
A process of obtaining the degree of coincidence of the semantic network and outputting a fixed message having the highest likelihood, and based on the output result, a part of the basic message similar to the fixed message is converted into a fixed message. Processing, extracting only active content that has meaning to a recipient who receives the message from the message, and outputting only active message corresponding to the active content in the converted boilerplate message And the activity that causes the computer to execute
The live message generator.

Unlike the automatic sentence summarizing device, the active message generating device of the present invention extracts a part related to active contents in a message sentence, and converts a fixed expression into a fixed sentence held by the system. This allows the message recipient to generate the most suitable active message.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, embodiments of the present invention will be described with reference to the drawings.

Referring to FIG. 1, an active message generating apparatus according to an embodiment of the present invention includes a basic message input unit 1 that controls an interface for inputting a basic message M.
And a basic message analysis unit 2 that analyzes the basic message M input by the basic message input unit 1 in a natural language, and the contents of the basic message M analyzed by the basic message analysis unit 2 and comparison data of existing text. , A message comparison unit 3 that outputs a comparison result indicating whether or not they are similar to each other, a comparison result of the message comparison unit 3 is acquired, and a message portion similar to the standard message is extracted from the basic message M, Based on the analysis result of the basic message M by the message standardization unit 4 and the basic message analysis unit 2 for converting into the fixed message, an active message portion in the basic message M that has meaning to the recipient who receives the message. Only the active message extractor 5 that extracts only the output of the message stylization unit 4 and the active message Comparing the output of the di-extraction unit 5, in the converted fixed message is configured in the active message output unit 6 is an interface portion for outputting an active message A corresponding to active content.

Next, the processing of each unit will be described in detail with a concrete example.

The basic message input unit 1 controls an interface for inputting the basic message M. The basic message M has various formats. The simplest form is an example where there is only BODY of the message (FIG. 2 (1)).
There is also an example (FIG. 2 (2)) in which a message bibliographic item is described and there is a message BODY related to it.

In the electronic mail, in addition to the above example, there are bibliographic items as shown in FIG. The meaning of the bibliographic items in the e-mail is shown on the left side after "#".

In the electronic mail, the name of the destination has a rule as shown in FIG. A name for identifying an appropriate user (for example, "taro") is described in the place of the user name. Of course, the use of Kanji code is not allowed, and 7
It is described in IA-5 (ASCII) represented by bits. Similarly, the machine name is also displayed hierarchically, and "nnn.
co. A domain name such as “jp” (jp means Japan, co means a company, and nnn means one of the companies) is described.
Although t is not specified in the e-mail, it may indicate the outline of the communication message, or may be given a classification name for classifying the content of the communication message. For example,
“Subject: [General] About Today's Testing”.
However, normally, in an e-mail, the Subject itself is displayed in ASCII as well as the communication destination name, or M
Binaryized Chinese characters are embedded by a special identifier called IME. Here, it is written in Kanji for easy reading.

The basic message analysis unit 2 analyzes the basic message M. First, the morphological analysis dictionary adds a part of speech, reading, and the like, and the morphological analysis result with the highest likelihood is extracted.

FIG. 5 shows an example 1 of the basic message M (see FIG. 2).
It is an example of morphological analysis of (1)). As can be seen from the example of FIG. 5, in the morphological analysis, the morphological candidate of the basic message M is extracted by the morphological analysis dictionary. For example, the candidates shown in FIG. 6 are extracted. In the case of "test", describe the morpheme. The part of speech for the morpheme will be described later in parentheses.
In the case of "test", all combinations of "test" and "no" are morpheme candidates (FIG. 7).

Similar morpheme candidate extraction is performed for all input characters. Finally, after all the morpheme candidates have been extracted, the link cost between the morpheme candidates (a measure representing the possibility of connection; here, the larger the cost, the lower the possibility of connection). Each link cost is determined for each part of speech or each morpheme. FIG. 8 shows an example.

Further, a node cost (a scale indicating the degree of use as a kanji of each word; here, the larger the number, the less used) is extracted from each word (FIG. 9).

Here, the cost of the morphological analysis candidate is expressed by the following equation.

[0020]

[Equation 1] The candidate having the smallest sum of the node cost and the link cost of the morpheme analysis candidates is the morpheme analysis candidate with high likelihood. Moreover, this morphological analysis method is called a cost minimum method.

FIG. 10 shows the cost when only the "test" portion of the example sentence of the basic message M is obtained.

Finally, the combinations of morpheme candidates are arranged in ascending order of cost, and the candidate with the lowest cost is taken out as the morpheme analysis result. The above example is rearranged in ascending order of cost of morpheme analysis candidates, as shown in FIG.

Finally, the combinations of morpheme candidates are arranged in ascending order of cost, and the candidate with the lowest cost is taken out. In the above example, the combination of "test (sahen noun. Stem) + wo (case particle. Wo)" is the morphological analysis result.

From this morphological analysis result, dependency analysis is further performed. Dependency analysis is performed by the patent “Dependency Analysis Method” filed by Inagaki et al. (Japanese Patent Application No. 62-17301).
1) etc. are used.

Example 1 of the basic message M is the dependency analysis as shown in FIG.

When Example 2 of the basic message M is shown in FIG. 13, Example 2 of the basic message M is the dependency analysis as shown in FIG.

The message comparison unit 3 compares the fixed text already held on the device side with the basic message M analyzed by the basic message analysis unit 2 to determine whether or not they are similar. The standard message depends on the field in which the message is used. For example, in the case of a message exchange service provided by PHS, mobile phones, etc., a message as shown in FIG. 15 can be considered as a standard message.

There are also short meaning symbols that represent a certain meaning, and these are also treated in the same way as fixed phrases. For example,
In the case of a message exchange service provided by PHS, mobile phones, etc., the meaning symbols as shown in FIG. 16 are conceivable.

Before these are processed by the message comparison unit 3, they are analyzed automatically or manually. In the message analysis, semantic analysis is performed from the dependency analysis analyzed by the basic message analysis unit 2, and the message is configured by a network having a semantic element and a role of the semantic element. A semantic element is a network representation of a representative meaning of each word and each combination (herein, they are called a semantic network). As a method of automatically performing the semantic analysis, a "pattern-guided document content analysis device" invented by Inagaki et al. (Japanese Patent Laid-Open No. 7-129588) is used.
For example, the basic message sentences (3-1) (3-2) (3-3) are as shown in FIG. 17, and the role of the semantic element is enclosed in a box. The role of semantics is agent, time, location, ac
t, object, etc. (4-1) etc. are basically
Converted to a semantic network. FIG. 18 shows an example thereof. When there are a plurality of semantic elements as in (4-3), a plurality of semantic element networks are prepared for fixed phrases.

In this way, the fixed texts already prepared on the device side are stored as a semantic network on the device side. On the other hand, here, the basic message output from the basic message analysis unit 2 is subjected to semantic analysis, and converted into a similar semantic network. For example, the basic message sentences (2-3) and (2-4) become a semantic network as shown in FIG.

In this way, after the semantics of the basic message M are analyzed, they are compared with the fixed form sentence. In the case of comparison, it is determined that the semantic network satisfying the semantic network of the standard sentence is the same. That is, if the two semantics networks to be compared have the same network shape and there is even one element of the content of the semantics at the network end, it is determined that they match. If they match, the degree of matching between the two semantic networks that are the targets of comparison of the sufficiency (a numerical representation of the network shape and the matching of the semantics at the network ends) is set, and the highest likelihood ( The ones with a high degree of matching) are output. For example, in Example 2 of the basic message M, as shown in FIG.
As shown in 0, (2-4) and the fixed sentence (3-3) have the highest degree of sufficiency of the semantic network. In the basic message M of (2), the information as shown in FIG. 21 is sent from the message comparison unit 3.

The message stylization unit 4 receives the output from the message comparison unit 2 and stylizes the basic message. In the example of the basic message M of (2) shown above,
The standard message part of the matching standard message is encoded and output. The part that does not match is output as it is as a free character. For example, if the prefix of the code (the meaning of the prefix; if there is a prefix, then two numbers represent a specific message, for example, here is the code of a fixed message) is “#”, (2 −
The message of 4) is as shown in FIG. Do the same for all basic message texts.

The active message extractor 5 is a part for extracting only active messages in the basic message sentence. As a method for active analysis of a basic message, there is an "event analyzer" (Japanese Patent Application No. 3-261306) invented by Inagaki et al. It is a device that selects only the active messages in a sentence and analyzes their meaning. In the "event analyzer", first, a target input sentence is divided into short sentence units (units of sentences having one active message in one sentence), and each short sentence is subjected to natural language analysis. It is determined whether or not the sentence is an active sentence by using the morphological analysis result in units of short sentences, and in the case of the active sentence, the semantic structure of the sentence is acquired. An example 2 of the basic message M will be described.

In the basic message, (2-3) has a plurality of active messages (a message of "meal" and two messages of "return"), so it is divided into two short sentences. (The beginning of the short sentence is “<sentence>” and the end is </ sentence>
Indicate. ). Example 2 of the basic message M is in short sentence units as shown in FIG. In the example of this basic message M,
The only active messages are (2-3) parts 1, 2 and (2-4). Therefore, only the sentence of FIG. 24 is extracted from the basic message M.

Further, the above sentence is analyzed by the active analysis, and the semantic structure shown in FIG. 25 is acquired.

The active message output unit 6 is an interface that combines the outputs of the message stylization unit 4 and the active message extraction unit 5 and outputs the combined output to the final output device. In the case of the example 2 of the basic message M, the output as shown in FIG.
Is output from. Furthermore, the active message extraction unit 5 outputs a message as shown in FIG.

The meaning display is extracted from the outputs of the message stylization unit 4 and the active message extraction unit 5 according to the following rules.

-If there is no <sem> tag in the <sentence> tag, the character string in the <sentence> tag is output.

-If there is a <sem> tag, the value of the value attribute of the <sem> tag is connected and output.

-If there is a <sem> tag, the character string in the <sem> tag is output.

-In the <sentence> tag, if num is the same,
Output all possibilities.

-When the value of the value attribute of the <sem> tag is shown in parentheses, when it is the same as when it is displayed on the output device,
Do not display the same content.

-Connect by "+" for each tag output unit.

The division of <sentence> unit is "x".

According to the above rule, Example 2 of the basic message M is as shown in FIG.

Although an example of candidates for each line has been shown, a candidate for inserting a line feed position between all words is created and scored. For example, as shown in FIG. As a scoring method, the optimum degree of display on the output device is represented by points. If it is suitable for display on the output device, point addition is performed, and if it is not suitable for display on the output device, point subtraction (penalty) is performed.

FIG. 30 shows an example of items for calculating points.

In the case of a candidate that matches each item, an operation of adding the point is performed, and it is compared whether all the items match, and each text conversion candidate is pointed. The item “text conversion candidate line * column considering line break position” means that the occupied character string on the screen occupied by the text conversion candidate considering line break position is the same as the screen size (column). Is less than the number of characters that can be displayed on one screen.

If the final output device can receive the fixed phrase code with the prefix, it is transmitted to the output device with the prefixed code. When the prefix is not recognized and only the 2-byte character is recognized, the fixed phrase is also converted into the 2-byte code and output. Of course, the attributes of the output devices also have the attributes peculiar to the respective output devices, so that the display most suitable for the output device must be performed. The attributes of the output device include the screen size (row x column) of the device to which the text is output, the number of characters that can be displayed on one screen, the maximum number of characters that can be displayed as a whole, and the like.

The attributes of the output device are as shown in FIG.

The number of characters is the same as the number of bytes. When displaying Japanese Kanji, the area of 2 bytes (2 characters) is always required. The screen size (row) is the number of resolution characters in the vertical direction of the screen of the output device (the character here is A
SCII 1-byte character). That is, in the above example, the number of characters that can be displayed simultaneously in the vertical direction is 2
It will be a character. What is the screen size (column)?
The horizontal resolution of the screen of the output device expressed in characters. When the attribute value has only one number, all lines are displayed by the number of characters indicated by the screen size (column). If a number of 1 or more is given as an attribute value, it represents the screen size (column) of each row. The number of characters that can be displayed on one screen is usually
Screen size (row) x screen size (column). The total maximum number of characters is the maximum number of characters that can be sent to the output device at one time as a communication message, and the following relationships are usually established.

(Total maximum number of characters) ≧ (number of characters that can be displayed on one screen) These output device attributes may be determined for each series of output devices or may be determined by a communication partner. In addition, the communication partner may have a plurality of output devices, and the attributes of the output devices may be undefined. In that case, the attribute of the output device is obtained with respect to the server (the output device may be the server directly). The server name describes the server name (may be a URL) that stores the output device attributes. The protocol means a client / server communication protocol for acquiring the output device attribute from the server. For example, HTTP, FTP,
There is TELNET etc. The description language part describes the structure of the description format in which the attribute is described when the attribute is acquired by the communication protocol. HTML and XML are possible. For example, when the above example is described in XML, the attribute information is described as shown in FIG. "<", ">" Are symbols that indicate the start and end of the tag. The tag name (here, the attribute name is described in the start and end tags. “<//”, “>”)
Indicates the end of the parameter indicated by the tag name. The attribute value of each attribute is described as a parameter between the above tags.

As described above, the item of each point is calculated, and the candidate with the highest point is output as the active message A as the maximum likelihood candidate.

The candidates output by the active message extraction unit 5 are sent by an appropriate protocol according to the transmission protocol of the output device. For example, if the protocol is HTTP, information is exchanged with the output device by HTTP, and the text conversion candidate is output.

For example, in Example 2 of the basic message M,
"Restaurant" + "Meals" + "NOW" X / "12: 0
0 ”+“ return ”+“? ” In case of # 33 ”+ /” after “X /”,
If it is HTTP, it communicates with the output device as shown in FIG.

← The output device transfers information from the output device to the device side.

→ The output device transfers information from the device side to the output device.

In HTML3.0, the line feed is "<BR>"
It is represented by.

As a result, the HTML is displayed as shown in FIG. 34 on the screen of the output device.

Referring to FIG. 35, an active message generating device according to another embodiment of the present invention is an input device 11 such as a keyboard and an output device 12 such as a CRT display.
And recorded the active message generator,
Recording medium 13 such as FD, CD-ROM, semiconductor memory
And a data processing device 14 for reading and executing the active message generation program from the recording medium 13.

Here, the active message generation program recorded in the recording medium 13 is, as shown in FIG. 36, the basic message processing 21 for inputting the basic message M from the input device 11 and the natural language analysis of the input basic message. A message that compares the basic message analyzed by the basic message analysis processing 22 with the standard message with the standard message, and indicates whether or not they are semantically similar, or outputs a comparison result that is the degree of similarity. A process 23 and a message stylization process 24 for converting a basic message similar to the standard message in the basic message M to a standard message based on the comparison result.
In the basic message M, only the active message extraction process 25 for extracting only active contents that have meaning to the recipient who receives the message, and only the active message A corresponding to the active contents in the converted standard message. Of the active message output process 26 for outputting the
The same process as that of 6 is performed.

[0062]

As described above, according to the present invention,
For message texts such as e-mail and TELEX,
By analyzing the message contents, converting the canonical expressions into canonical expressions, and extracting only the active messages in the message, the message contents are less redundant to the message recipient and the most informative. It can be properly displayed in a large amount.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an active message generation device according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a basic message.

FIG. 3 is a diagram showing bibliographic items in an electronic mail.

FIG. 4 is a diagram showing a rule of a destination name in an electronic mail.

5 is a diagram showing a result of morphological analysis of an example of a basic message M. FIG.

FIG. 6 is a diagram showing extracted morpheme candidates.

FIG. 7 is a diagram showing morpheme candidates in the case of “test”.

FIG. 8 is a diagram showing an example of link costs.

FIG. 9 is a diagram showing a node cost extracted for each word.

FIG. 10 is a diagram showing costs of only a “test” portion.

11 is a diagram in which the combinations of morphemes in FIG. 10 are arranged in ascending order of cost.

12 is a diagram showing a dependency analysis result of an example of a basic message M. FIG.

FIG. 13 is a diagram showing a second example of a basic message M.

14 is a diagram showing the dependency analysis result of Example 2 of the basic message M. FIG.

FIG. 15 is a diagram showing a fixed message text in the case of a message exchange service.

FIG. 16 is a diagram showing the meaning of symbols in the case of a message exchange service.

FIG. 17 is a diagram showing a semantic network of basic message sentences of (3-1), (3-2), and (3-3).

FIG. 18 is a diagram showing a semantic element network of basic message sentences of (4-1) and (4-2).

19 is a diagram showing a semantic network of basic message sentences of (2-3) and (2-4). FIG.

FIG. 20 is a diagram showing that the degree of sufficiency of the semantic network of the basic message sentences (2-4) and (3-3) is the highest.

FIG. 21 is a diagram showing information sent from the message comparison unit 3.

FIG. 22 is a diagram showing an example of a stylized message of a basic message (2-4).

FIG. 23 is a diagram in which Example 2 of the basic message sentence M is divided into short sentence units.

24 is a diagram showing a sentence extracted from a basic message M. FIG.

FIG. 25 is a diagram showing a semantic structure acquired by active analysis.

FIG. 26 is a diagram showing a message output from the message stylization unit 4.

FIG. 27 is a diagram showing a second example of the basic message M from which the meaning display is extracted.

28 is a diagram showing a result of extracting a meaning display from Example 2 of the basic message M. FIG.

FIG. 29 is a diagram showing a result of scoring the example of FIG. 2.

FIG. 30 is a diagram showing an example of items for calculating points.

FIG. 31 is a diagram showing attributes of an output device.

FIG. 32 is a diagram showing description information when an attribute is described in XML.

FIG. 33 is a diagram showing communication between an output device and an apparatus.

FIG. 34 is a diagram showing a display example on the screen of the output device.

FIG. 35 is a configuration diagram of an active message generation device according to another embodiment of the present invention.

36 is a configuration diagram of an active message generation program recorded in the recording medium 13 in FIG. 34.

[Explanation of symbols]

1 Basic message input section 2 Basic message analysis section 3 Message comparison section 4 Message stylization part 5 Active message extractor 6 Activi message output section 11 Input device 12 Output device 13 recording media 14 Data processing device 21 Basic message input processing 22 Basic message analysis processing 23 Message comparison processing 24 Message standardization processing 25 Active message extraction processing 26 Active message output processing

Front page continuation (72) Inventor Kazuo Tanaka 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo Nihon Telegraph and Telephone Corporation (56) Reference JP-A-5-73611 (JP, A) JP-A-9 -205664 (JP, A) JP 7-129588 (JP, A) JP 63-175965 (JP, A) JP 64-17152 (JP, A) (58) Fields investigated (Int.Cl) . ⁷ , DB name) G06F 17/21-17/28 G06F 13/00 JISST file (JOIS)

Claims (7)

(57) [Claims] 1. Receiving a message within a message
Only active messages that are meaningful to the recipient
Which is a process in an apparatus for generating a
A number of standard messages composed of
It is and performs the steps of inputting a basic message, a step of morphological analysis the input base message, the dependency analysis of the morphological analysis result, the basic message
A sage is a semantic network consisting of semantics and roles of semantics.
And the semantic network of the basic message is a standard message.
When satisfying the semantic network of
It is judged that this message matches the standard message, and
If there is a match, the degree of coincidence of the semantic network is calculated.
Because, and outputting the highest likelihood fixed message, on the basis of the output result, and converting the fixed message parts that are similar to the fixed message in the base message, the basic message In which only active content that is meaningful to the recipient receiving the message is extracted, and in the converted basic message, only active messages corresponding to the active content are output. Message generation method. 2. A basic message input means for inputting a basic message, and a form for all characters of the input basic message.
Morphological candidates are extracted using the dictionary for morphological analysis,
After the candidate elements are extracted, the degree of use of each form
Node cost of each morpheme shown and connection between each morpheme candidate
Find the link cost, which is a measure of continuity
Node cost and phosphorus for combinations of morpheme candidates
In order of lowest total cost, the lowest cost
Outputs the complement as a morphological analysis result,
Dependency analysis is performed, and the basic message is regarded as a semantic element.
A basic message analysis means composed of a semantic network consisting of the roles of semantics and a fixed message composed of a semantic network,
A standard message that stores a semantic network of standard messages.
The message storage means and the semantic network of the basic message are standard messages.
When satisfying the semantic network of
It is judged that this message matches the standard message, and
If there is a match, the degree of coincidence of the semantic network is calculated.
Because the message comparison means for outputting a highest likelihood fixed message, based on a result of the message comparison means, the fixed message similar to fixed message part <br/> are among the basic message Message formalizing means for converting, active message extracting means for extracting only active content that has meaning to a recipient who receives the message in the basic message, and active content in the converted basic message And an active message output means for outputting only an active message corresponding to the above. Wherein the message stylized unit outputs encoding the regular message portion mismatch sense element, according to claim 2, wherein. 4. The active message extracting means,
The basic message sentence is divided into short sentence units, which are units of a sentence in which one active message is included in one sentence, each short sentence is subjected to natural language analysis, and an active sentence is obtained using a morphological analysis result of each short sentence unit. To determine if
The apparatus according to claim 2 , wherein in the case of an active sentence, its semantic structure is obtained. 5. The active message output means comprises:
3. The apparatus according to claim 2, wherein the active message is output to the output device with the least number of codes based on the code table describing the notation suitable for indicating the meaning to be output. 6. The active message output means comprises:
The device according to claim 2, wherein the text conversion candidates are output by an appropriate protocol according to the transmission protocol of the output device. 7. A process of inputting a basic message, a process of morphological analysis of the input basic message, and a dependency analysis based on a morphological analysis result, and the basic message having a meaning consisting of a semantic element and a role of a semantic element. The process of configuring a basic network, the process of reading a plurality of standard messages composed of a semantic network, and when the basic message's semantic network satisfies the standard message's semantic network, the basic message is It is determined that the message matches, and if they match, the degree of matching of the semantic network is calculated, and the process of outputting the most likely fixed message, based on the output result, the fixed message in the basic message. The process of converting a part similar to Computer , a process of extracting only active contents having meaning to a recipient who receives the message, and a process of outputting only active messages corresponding to the active contents in the converted standard message. Computer readable recording active message generator for execution by
Capacity recording medium.