JP5573257B2 - Character input program and communication terminal device - Google Patents

Description

  The present invention provides a program for causing a computer having a function of transmitting and receiving an e-mail to execute a process of inputting a character string corresponding to the operation to a mail to be transmitted according to an operation of inputting characters, The present invention also relates to a communication terminal device using this program.

  For mobile phones in Japan, conversion candidates corresponding to reading at the time of input each time an operation is performed to input kana characters that make up the reading of the characters to be input, in order to eliminate the complexity of character input Is set (hereinafter, this function is referred to as “predictive conversion function”). This predictive conversion function allows the user to select and confirm the conversion candidate when the target character string is displayed as a conversion candidate without inputting all the readings. The effort is greatly reduced.

In addition, when creating a reply email for a received email as a model that has improved convenience, a response corresponding to the word included in the received email is entered, Have a function of displaying the corresponding word as an upper conversion candidate. Hereinafter, this function is referred to as “reply automatic learning function”.
The reply automatic learning function is effective only when a reply mail for the received mail is created.

For example, Patent Document 1 exists as a prior art document related to the automatic reply learning function.
In Patent Literature 1, when an instruction to create a reply mail based on the received mail is given, a word is extracted from the received mail, and if an unregistered word is extracted, the word is used. If you register with a frequency of 0 and an already registered word is extracted, set the usage frequency exception flag for that word, or reply only based on the received mail. Among the extracted conversion candidates, it is described that the one with a usage frequency of 0 or the one with the usage frequency exception flag set is displayed at the top.

  Further, Patent Document 2 discloses an invention in which even a character string that is not described in a received mail displays a character string that is likely to be input from the contents of the received mail. Has been. Specifically, this Patent Document 2 classifies words in a dictionary into a plurality of categories, and when receiving a reply mail creation instruction for the received mail, analyzes the received mail and classifies the words corresponding to the words in the document. It is described that the word corresponding to the classification specified in the processing for the received mail is preferentially displayed with respect to the input of reading. With this process, for example, when a reply mail is created for a received mail containing a word that belongs to the place name "Osaka", if a reading of "U" is entered, the place name such as "Umeda" Can be displayed.

JP 2006-344039 A JP 2007-94786 A

The automatic reply learning function is a highly convenient function, but this function is not always useful. An example of this will be described with reference to FIG.
FIG. 15 shows an example of the received mail, and as a reply mail creation screen for the received mail, a screen (A) to which the automatic reply learning function and the predictive conversion function are applied, and a screen to which only the predictive conversion function is applied (B) is illustrated. In either example, in response to the question “Would you like to go to cherry blossom viewing?” In the received mail, a reading “I” is entered, and conversion candidates corresponding to the reading are displayed. As is clear from the comparison of both screens and the contents of the received mail, among the conversion candidates on screen (A), the first “Do not go” and the second “From now” reply automatic learning function The conversion candidates of the third and lower ranks are extracted by the predictive conversion function.

Normally, when you answer a question that includes a usage word, such as “Would you like to go?” In the example above, like “I will go” or “I will not go”. A different wording from the question should be used.
However, in the general automatic reply learning function, the character string of the phrase unit included in the received mail is registered as it is, so even a word including a utilization form is also displayed in the received mail as shown in the screen (A). The included expression “Would you like to go?” Is displayed as it is. The screen (A) in FIG. 15 also displays a conversion candidate “go”. This display is based on the predictive conversion function. In candidate extraction by the predictive conversion function, since the display order of conversion candidates is determined based on the past use frequency, when the use frequency of “go” is not high, it is not preferentially displayed. In the illustrated example, candidates that are likely to be used as a response such as “Let's go” and “I will go” are not displayed, so it takes time to call the target candidate.

  The present invention pays attention to the above-mentioned problems, and when using the automatic reply learning function in response to the reception of the mail including the usage word, the expression using the usage word in the received mail is replaced with another expression. It is an object of the present invention to greatly improve the convenience of the automatic reply learning function by displaying things preferentially as conversion candidates.

The program according to the present invention converts a computer having a function of transmitting and receiving an e-mail into the following conversion rule storage means, morpheme analysis means, converted character string storage means, learning processing means, candidate extraction means, candidate display means. The above problem is solved by functioning as an input character string determination means.

The conversion rule storage means, for multiple types of usage patterns, is a conversion rule for replacing the expression in the usage form of the pattern with another expression, and the usage form after replacement with the combination of the usage form before replacement and the attached word. A conversion rule having a configuration in which a combination of an adjunct is associated is stored, and a post-conversion character string created by applying the conversion rule is stored in the post-conversion character string storage means . The morpheme analysis means executes morpheme analysis for one of the received mails. The learning processing means causes the morpheme analysis means to perform a morpheme analysis on the reply target mail in response to an operation for instructing a reply to one of the received mails, and then extracts by this morpheme analysis process for processing the clause for each clause comprising utilizing word that is, utilization pattern of use words in a string to be processed clauses and the conversion rule storing means by the combination with the accompanying word in its conjugations processed clause Search and extract the conversion rule that matches the combination, and apply the combination of the replacement usage type and the attached word set in the extracted conversion rule to the processing target clause, The step to be created is executed for each clause to be processed. Then, the created converted character string is made valid only while the reply mail creation screen for the operation for instructing the reply is activated, and is associated with the reading of the converted character string. Save in the character string storage means after conversion .

The candidate extraction unit is configured to store the post-conversion character string storage unit when an operation of inputting a reading of a character string to be input is performed on a reply mail creation screen launched in response to an operation to instruct a reply. A search using the input reading is executed on the converted character string stored in a valid state, and a converted character string that matches the reading is extracted as a conversion candidate. The candidate display means displays a candidate display field including the conversion candidate extracted by the candidate extracting means on the reply mail creation screen.
The input character string confirming means confirms the selected conversion candidate character string as an input character string to the reply mail in response to an operation of selecting one of the conversion candidates displayed in the candidate display field.

Of the two types of storage means described above, the conversion rule storage means is mainly a data file in which electronic data indicating the conversion rule is stored. This storage means is stored in the memory of a computer where the program is installed. The converted character string storage means is also set in the memory of the computer where the program is installed, but no data is stored at the time of installation, and processing by the learning processing means is performed to create a converted character string. When converted, the converted character string is stored together with the reading.

  “Utilization pattern” refers to a pattern of change in each utilization form of utilization words. For example, with regard to verbs, there are utilization patterns such as five-stage utilization, lower-stage utilization, upper-stage utilization, ka-line modification utilization, sa-line modification utilization, and the like. In the case of a verb, in any utilization pattern, conversion rules can be set for each of the utilization forms of the green form, the continuous form, the end form, the combined form, the hypothesis form, and the instruction form.

  As conversion rules, in addition to rules for replacing an expression using a certain usage form with an expression using another usage form, it is possible to replace only an ancillary word without changing the usage form. You can also set rules.

According to the computer in which the above program is installed, the user performs an operation to instruct a reply to the received mail in response to receiving the mail including the used word, and the used word is obtained from the received mail by the morphological analysis means. When the included phrase is extracted, processing for converting the processing target phrase into a character string representing another expression is executed by the learning processing unit with this clause as a processing target. The converted character string generated by this conversion processing is stored in the converted character string storage unit in association with the reading of the character string. Furthermore, when a reading corresponding to the converted character string is input to the reply mail creation screen, the converted character string corresponding to the reading is extracted as a conversion candidate from the converted character string storage means, and is displayed in the candidate display field. Is displayed. Further, when an operation for selecting the displayed converted character string is performed, the converted character string is determined as an input character string and input to the reply mail.

  Therefore, if a sufficient number of conversion rules are set for each usage pattern and stored in the conversion rule storage means, the phrases containing the usage words in the received mail are replaced with character strings of various expressions. Can be displayed as conversion candidates. Therefore, when the user tries to input a character string having a different expression from the received mail using the same language as the received mail and inputs the reading of the character string, the target character string is displayed as a candidate for conversion as the target character string. Can be displayed.

In a preferred aspect of the program according to the present invention, the learning processing means executes processing for each processing target clause after clearing the converted character string storage means in response to an operation for instructing a reply to one of the received mails. The candidate extracting unit executes a search for the converted character string storage unit after being cleared in response to the operation for instructing the reply .

According to the computer in which the program according to the present invention is introduced, it is possible to easily convert a character string based on a combination of a predetermined usage word and an attached word into a character string indicating another expression using the same usage word. Is possible.

  Furthermore, in another preferred aspect of the program according to the present invention, each conversion rule stored by the conversion rule storage means includes information indicating the use frequency of the conversion rule. Also, when the converted character string displayed as the conversion candidate by the candidate display means is confirmed as the input character string to the reply mail by the input character string confirmation means, the conversion used for the conversion to the confirmed character string Further included is a program that causes a computer to function as a usage frequency update means for updating the usage frequency of a rule. Further, when a plurality of converted character strings are extracted by the candidate extracting means, the candidate display means determines the display order of these converted character strings by using the frequency of use of the conversion rules used for conversion into the respective character strings. Configured to be used.

  In a more preferred embodiment, each conversion rule stored in the conversion rule storage unit is linked to each conversion rule having a similar concept, and the usage frequency update unit converts the conversion rule to the character string determined by the input character string determination unit. The usage frequency of the conversion rule used for the conversion rule is updated together with the usage frequency of the conversion rule associated with the conversion rule.

  According to each of the above aspects, the use frequency of the conversion rule used for the conversion to the expression that the user prefers is increased, and the character string candidates converted by the conversion rule are displayed at the top. become. Therefore, it is possible to display conversion candidates that match the expression that the user likes and uses at the top, making it easier to perform operations for selecting candidates and further improving convenience in character input. Become.

A communication terminal device to which the present invention is applied includes a transmission / reception unit for transmitting and receiving an e-mail, an operation unit including a plurality of keys to which kana characters are assigned, and a display for displaying an e-mail document creation screen Part. Further, the communication terminal device includes a conversion rule storage means, a converted character string storage means, a morpheme analysis means, a learning processing means, a candidate extraction means, a candidate display means, and an input character string determination means according to the above program. Provided.
The communication terminal device is configured as a mobile communication device such as a mobile phone or a PDA. The personal computer also functions as a communication terminal device by incorporating the above program.

  According to the present invention, it is possible to easily input a character string in which an expression including a usage word in a received mail is replaced with another expression when creating a reply mail for a received mail including the usage word. Therefore, it is possible to greatly improve the convenience in character input. Therefore, even a communication terminal device with poor operability such as a mobile phone can efficiently create a reply mail.

It is a figure which shows the external appearance of the mobile telephone to which this invention was applied. It is a block diagram which shows the structure of a mobile telephone. It is a figure which shows the example of a received mail, and the example of a display of the conversion candidate in the preparation screen of the reply mail with respect to this received mail. It is a functional block diagram which shows the structure of a character input system. It is a figure which shows the data structural example of a wording replacement table. It is a figure which shows the data structural example of a utilization form table. It is a figure which shows the clause of the processing target of morphological analysis, and the result of the morphological analysis with respect to this clause. It is a figure which shows the specific example of the conversion process with respect to the phrase containing a utilization word. It is a figure which shows the specific example of the conversion process with respect to the phrase containing a utilization word. It is a flowchart which shows the schematic procedure of a reply automatic learning process. FIG. 11 is a flowchart showing a detailed procedure related to a dictionary data creation process (step S5) in the process of FIG. It is a flowchart which shows the detailed procedure regarding a conversion process (step S16) among the processes of FIG. It is a figure which shows the other structural example of a wording replacement table. It is a flowchart which shows the procedure of the update process of the usage frequency with respect to the wording replacement table of FIG. It is a figure which shows the example of a display of the conversion candidate in the creation screen of the reply mail using the conventional reply automatic learning function and the prediction conversion function.

FIG. 1 shows the appearance of a mobile phone as an application example of the present invention, and FIG. 2 shows the electrical configuration of the mobile phone.
The mobile phone 1 of this embodiment has a configuration in which a camera 4, an image processor 5, a communication circuit 6, a control circuit 10, and the like are accommodated in a housing 1A provided with a display unit 2 and an operation unit 3. In addition to calling, it has functions to connect to the Internet and send and receive e-mails. It is also possible to execute imaging by the camera 4 under the control of the image processor 5 and output the generated image to an external device such as a personal computer, or transmit the generated image attached to an e-mail.

Control circuit 10, in addition to the CPU 11, main memory 12, and the like work memory 13. The main memory 12 is a non-volatile memory such as a flash memory, and stores programs, setting data, and the like. On the other hand, the work memory 13 is a volatile memory (RAM), and stores conversion candidates and automatic learning dictionaries extracted by character input processing described later.

  The display unit 2 is composed of a liquid crystal panel. The operation unit 3 is provided with a plurality of keys including a character key 31 and a cross key 32 to which numbers “0” to “9” are assigned. In addition to numbers, each character key 31 is assigned a hiragana character of one line of 50 sounds or a plurality of alphabet characters or symbols. In addition, the structure of the operation part 3 is not restricted above, For example, you may make a touch panel display which displayed the movable keyboard function as an operation part.

  When a Japanese input mode is set, such as when composing an email, while the same character key 31 is being pressed continuously, the kana characters assigned to that key are called in order, and other key operations are performed. A kana character string representing the reading (hereinafter referred to as “reading character string”) is assembled by a method of determining the kana character called immediately before according to the above. In addition, when conversion candidates are extracted and displayed in conjunction with changes in the reading character string and an operation for selecting one of the displayed conversion candidates is performed, the character string of the selected candidate is the input character. Confirmed as a column.

  Furthermore, the mobile phone of this embodiment has a built-in automatic reply learning function, and when an email for replying to a received email is created, when reading of an independent word included in the received email is input, A conversion candidate including the corresponding independent word is preferentially displayed. In particular, in this embodiment, when a reading corresponding to a usage word in a received mail is input, a character string having a configuration in which the corresponding usage word is converted into an expression different from that of the received mail is displayed as a conversion candidate. be able to.

  FIG. 3 shows an example in which a mail containing the phrase “Do you want to run?” Containing the verb continuous form is received and a reply mail for the received mail is created. In the reply mail creation screen in this example, a candidate display column 20 is activated in response to the input of “ha”, and a plurality of conversion candidates whose readings begin with “ha” are displayed. Among the candidates displayed in the candidate display column 20, there are six candidates derived from the phrase “Do you want to run?” In the received mail (“Run”, “Do not run”, “Run”, “Run”) “Run” and “do not run”) and candidates extracted by the predictive conversion function (such as “fast” and “flower”).

FIG. 4 is a functional block diagram showing a system configuration related to the above character input.
The character input system 100 of this embodiment includes a control processing unit 101, a reading character string assembly unit 102, a candidate extraction unit 103, a reply automatic learning processing unit 104, a morpheme analysis unit 105, and a processing unit, a prediction conversion dictionary 111, It is composed of a storage unit such as an automatic reply learning dictionary 112, a morphological analysis dictionary 113, a word replacement table 114, and a utilization table 115. Each storage unit is constituted by a data file stored in the main memory 12. The entity of the processing unit is the CPU 11 that executes a program stored in the main memory 12.

In the figure, the mailer 200 shown in the upper left performs transmission and reception of mail in cooperation with the communication circuit 6 shown in FIG. 2 , creation of mail to be transmitted, browsing of received mail, reception folder and transmission folder. Perform management. The character input system 100 is activated in response to an operation for creating a reply mail for a received mail or an operation for instructing creation of a new mail, and a character string to be read in accordance with an operation performed on the operation unit 3 . The process from assembly to confirmation of the input character string is executed, and the confirmed input character string is output to the mailer 200.

The morpheme analysis dictionary 113 stores information necessary for resolving text data into units of words and recognizing the part of speech and usage of each word. A plurality of words are registered in the predictive conversion dictionary 111 in association with the readings.
In FIG. 4, one dictionary file for the morphological analysis dictionary 113 and one for the predictive conversion dictionary 111 is shown. However, the present invention is not limited to this, and any dictionary can be composed of a plurality of dictionary files.

The automatic reply learning dictionary 112 is for storing dictionary data created by the automatic reply learning function, and is validated only when a reply mail for the received mail is created. Further, the dictionary data stored in the reply automatic learning dictionary 112 is intended to be used only for certain reply mail, after the reply mail is created, the creation of the next reply mail is instructed Cleared accordingly.

  The wording replacement table 114 and the utilization form table 115 store conversion rules for replacing expressions (expressions) by various utilization words with other expressions by the same utilization words. FIG. 5 shows a data configuration example of the wording replacement table, and FIG. 6 shows a data configuration example of the utilization type table.

With reference to FIG. 4 again, the basic operation of the character input system 100 will be described. The control processing unit 101 performs character display on the display unit 2 and character input processing for a mail being created while receiving various key operations and communicating with other processing units.
First, when the character key 31 to which the kana character is assigned is operated, the control processing unit 101 informs the reading character string assembling unit 102 which key has been operated to assemble the reading character string. The read character string is displayed on the display unit 2.

  In addition, the control processing unit 101 starts up the candidate display field 20, and whenever the reading character string changes due to the input of a new kana character, the changed reading character string is passed to the candidate extraction unit 103 to instruct conversion. . In response to this instruction, the candidate extraction unit 103 searches the predictive conversion dictionary 111 and extracts conversion candidates having a reading that matches the reading character string in the forward direction. Further, the candidate extraction unit 103 performs processing for determining the display order of the extracted conversion candidates, creates a list of conversion candidates in the determined order, and passes this to the control processing unit 101. The control processing unit 101 displays each conversion candidate in the candidate display column 20 based on this list.

  When an operation for creating a reply mail for the received mail is performed, the control processing unit 101 passes the text data of the received mail to the reply automatic learning processing unit 104, and dictionary data for automatic reply learning Instruct the creation of.

  Receiving the above instruction, the automatic reply learning processing unit 104 passes the text data of the received mail to the morpheme analysis unit 105 to execute morpheme analysis. The morpheme analysis unit 105 recognizes the text data by dividing it into phrase units using the morpheme analysis dictionary 113, and further recognizes each phrase separately into independent words and attached words. The reply automatic learning processing unit 104 creates dictionary data corresponding to each independent word in the received mail using the result of the morphological analysis, stores it in the reply automatic learning dictionary 112, and performs candidate extraction processing. The reply automatic learning dictionary 112 is set to be valid. In accordance with this setting, the candidate extraction unit 103 searches both the prediction conversion dictionary 111 and the automatic reply learning dictionary 112 to extract conversion candidates, and prioritizes the conversion candidates extracted from the automatic reply learning dictionary 112. And display.

  The reply auto-learning processing unit 104 creates dictionary data that associates readings with the extracted words or phrases containing the extracted words or words that are not used among the independent words extracted by the morphological analysis. On the other hand, for verbs such as verbs, the automatic reply learning processing unit 104 uses the word replacement table 114 and the usage table 115 to change the character string that constitutes the phrase including the usage word to the same usage word. Is converted into a character string representing the expression of, and dictionary data in which the converted character string and the reading are associated with each other is created.

In the following, the automatic reply learning process for the target words will be described in detail.
FIG. 7 shows the result of morphological analysis for the phrase “Would you like to run?” Included in the received mail shown in FIG. This clause is recognized as being divided into the independent word “run” and the ancillary word “isn't it?”. Furthermore, the independent word “run” is broken down into the stem “run” and the inflection ending “ri”, and the reading of the stem, the type of part of speech, the utilization pattern and the utilization form are also recognized.

In the morphological analysis of this embodiment, information (hereinafter referred to as “part of speech information”) in which the recognition result of the utilization pattern is combined with the recognition result of the part of speech is created. In addition, utilization forms are expressed in combination with utilization patterns.
For example, in the example of FIG. 7, the part-of-speech information “La line five-stage verb” indicating that the verb is a La line five-stage inflection is set as a recognition result for the independent word “run”. Information "5-stage continuous form" indicating that "5-stage continuous form" is set. In addition, information indicating the recognition result is set for the reading of the stem and the ending of the usage.

  The wording replacement table 114 shown in FIG. 5 includes a plurality of rules (hereinafter referred to as “replacement rules”) for replacing the expression using the utilization form with another expression for each utilization form belonging to various utilization patterns. Stored. The replacement rule in this example has a configuration in which a replacement pattern representing an expression after replacement, a standard priority, and a usage frequency are associated with a utilization form and an attached word pattern representing an expression before replacement. FIG. 5 shows a specific data structure of a replacement rule, taking an example of a verb continuous form utilizing five steps.

  Further, in the wording replacement table 114 of this embodiment, each replacement rule is managed by an ID number. Hereinafter, when individually referring to the seven replacement rules shown in FIG. 5, they are referred to as “replacement rule 1” and “replacement rule 2” using the respective ID numbers.

  An ancillary word pattern is a wording pattern of the head part of an ancillary word included in the expression before conversion. For example, in the replacement rule 1 and 2 in the figure, the attached word pattern "None ~" has an attached word preceded by "Not" ("No" "Is it not" "No" "" ”Is included), and the adjunct pattern“ Masho ~ ”of replacement rule 3 includes an adjunct (“ Masho ”,“ Makesho ”,“ Let ’s do it ”). Yo ”,“ Masho Ne ”, etc.). In addition, the case where the adjunct pattern is “arbitrary” includes not only the case where an arbitrary character string is set as an adjunct but also the case where no adjunct is set depending on the utilization form.

The replacement pattern indicates a combination pattern of utilization and ancillary words in the expression after conversion, but there is also a replacement pattern in which only the utilization form is defined, such as replacement rules 4-7.
The standard priority and use frequency are used when determining the display order of conversion candidates created based on the replacement rule. The standard priority is determined based on the general usage frequency of the replacement pattern, and the value of each standard priority is fixed in this embodiment. On the other hand, the use frequency is updated in accordance with the fact that dictionary data created by the corresponding replacement rule is displayed as a conversion candidate and actually selected.

  According to the example of FIG. 5, in the conversion process using the replacement pattern 1, the utilization form is the same continuous form as before conversion, and a character string in which only the attached words are changed is created. In the replacement patterns 2 and 3, the same character string as before conversion is created except that the end of the attached word pattern (the part of ˜) is deleted. In addition, according to the replacement patterns 4, 5, and 6, a character string having only an inflected form is created without an attached word.

  In the utilization form table 115 shown in FIG. 6, for each type of part-of-speech information, the reading of the stem of the utilization word in the expression before conversion, the utilization form after conversion, the reading of the stem after conversion, and the utilization ending after conversion are stored. A plurality of combinations of information are registered. The information for one record stored in the utilization form table 115 indicates a rule for deriving a specific notation by applying the utilization form defined in the replacement pattern to the converted character string. Hereinafter, this rule is referred to as “utilization setting rule”.

  Each utilization form setting rule of the utilization form table 115 is given an ID code by a combination of alphabets and numbers. As a specific example, FIG. 6 shows utilization form setting rules A1 to A5 related to verbs utilizing the five-stage ra line and utilization form setting rules B1 to B5 for verbs utilizing the K line modification (ka variation). As in the utilization form setting rules A1 to A5, when the end of the stem before conversion is set to “arbitrary”, the rule is applied without considering the reading at the end. In addition, if the stem reading after conversion is set to “No Conversion”, it means that the reading of the stem before conversion is maintained, and if the usage ending is “None”, the usage ending is added. Means no.

  Based on the result of the morphological analysis, the reply automatic learning processing unit 104 recognizes part-of-speech information and ancillary words of a phrase character string (hereinafter referred to as a “character string to be converted”) including a utilization word extracted from the received mail. The replacement rules that match these are read from the wording replacement table 114.

  Further, the automatic reply learning processing unit 104 uses the prevailing form part-of-speech information and stem reading, and the utilization form defined in the replacement pattern of the replacement rule extracted from the wording replacement table 114. The table 115 is searched, and a utilization type setting rule that matches the combination of these information is specified. And, by applying the reading of the stem of the converted character string defined in the specified usage rule and the usage ending to the usage word of the conversion target character string, a concrete notation of the usage is derived, A post-conversion character string is created by applying the definition of the substitution pattern appendix to this.

  FIG. 8 shows an example of a character string using the verb continuous form utilizing the five-stage La row, and gives the conversion target character string “Would you like to run?” Illustrated in FIG. The process in the case of converting into a character string by another expression using the wording replacement table 114 and the utilization type table 115 is shown. Hereinafter, the conversion process using the word replacement table 114 and the utilization type table 115 will be specifically described with reference to FIG. 8 together with FIGS.

  When the word replacement table 114 shown in FIG. 5 is searched by the combination of the utilization form of the conversion target character string “Would you like to run?” In FIG. 8 and the attached word, the replacement rules 1 and 2 are assumed to be suitable for this combination. , 4, 5, 6, and 7 are extracted.

The replacement rule 1 defines that an attached word “mass” is attached to a word in which a conjugation word of a conversion target character string is expressed in a continuous form. When a utilization form setting rule is retrieved for the utilization form table 115 shown in FIG. When this utilization form setting rule A2 is applied to the independent word “run” of the conversion target character string, the continuous form notation “run” is maintained as it is. Furthermore, based on the replacement pattern of the replacement rule 1, by adding the attached word “mass” to this notation, a character string “run” indicating an expression different from the conversion target character string is set.

  The replacement rule 2 defines that an attached word “not” is attached to a word in which a conjugation word of a conversion target character string is expressed in a continuous form. Also in this case, similarly to the replacement rule 1, the utilization form setting rule A2 is applied, and “run” is derived as the utilization form notation. Therefore, by adding the attached word “not” defined by the replacement pattern of replacement rule 2 to this notation, “does not run” is set as the converted character string.

  Next, the replacement rule 4 defines that the word to be used in the conversion target character string is expressed in a terminal form and no additional word is attached. When the utilization form setting rule in the case of expressing the verb of the fifth row utilization in the last form is searched for the utilization form table 115 of FIG. 6, rule A1 is hit. Therefore, based on this utilization form setting rule A1, “run” is derived by maintaining the reading of the stem of the utilization word “run” of the conversion target character string and changing the utilization ending to “ru”. Since an attached word is not defined in the replacement pattern of the replacement rule 4, this “run” is set as a converted character string.

Similarly, when the replacement rule 5 is applied, the utilization type setting rule A4 is applied, and the free-standing word “run” in the converted character string is converted into the combined form “run”. , This is set to the converted string. In addition, when the replacement rule 6 is applied, the utilization form setting rule A5 is applied, and the utilization word “run” in the converted character string is converted into the continuous form of the notation “run”, which is converted Set to the back string.
When the replacement rule 7 is applied, the utilization type setting rule A3 is applied, and “run” is derived as an unwritten form. Further, this notation is combined with the attached word “none” defined in the replacement pattern of the replacement rule 7 to set “do not run” as the converted character string.

FIG. 9 shows an example of conversion processing when the character string “Would you like to come?” Using the continuous form of the verb “Kuru”, which utilizes the Ka-line transformation, is used as the conversion target character string.
Although not shown in FIG. 5, the replacement rules having the same contents as the replacement rules 1 to 7 are also set in the continuous form of the verb for utilizing the Ka-line modification. In FIG. 9, the ID numbers of these replacement rules are 101 to 107, and the replacement rules 101, 102, 104, 105, 106, and 107 are extracted for the conversion target character string "Will you come?" And

According to the utilization form table 115 in FIG. 6, when the replacement rules 101 and 102 are applied, the utilization form “Ki” is derived based on the utilization form setting rule B2, and the attached word is added to this. The combined character strings “coming” and “not coming” are set in the converted character string. Also, when the replacement rule 104 is applied, the end type “coming” is derived based on the utilization type setting rule B1, and this is set as it is in the converted character string.

When the replacement rule 105 is applied, the notation “Kita” in the form of a combined form is derived based on the utilization type setting rule B4, and this is set in the converted character string. When the replacement rule 106 is applied, the continuous type “Come” is derived based on the utilization type setting rule B5, and this is set in the converted character string. When the replacement rule 107 is applied, the original form “K” is derived based on the utilization form setting rule B 3, and this is combined with “None” of the ancillary word defined in the replacement rule 107. The character string “Do not come” is set as the converted character string.

  As shown in the above examples, in this embodiment, a plurality of replacement rules are set for each combination of utilization pattern and utilization form, and utilization form setting rules for deriving specific notation for various utilization forms. Are defined individually, there is no risk of grammatical errors during conversion. Therefore, it is possible to set a plurality of dictionary data in a character string with an expression different from that of the received mail, so that it is possible to enhance the automatic reply learning function for the used words.

  In this embodiment, in the wording replacement table 114, only the usage form of the character string after replacement is defined, and the rules for deriving a specific notation of the usage form are stored in the usage form table 115. Not limited to this, all the rules may be integrated into the wording replacement table 114.

  Further, in this embodiment, the replacement rule to be used is limited based on the combination of the utilization form of the conversion target character string and the attached word, but this limitation range can be defined and changed, and the conversion target character The conversion process may be performed using all the replacement rules corresponding to the column utilization pattern. For example, although the replacement rule 3 is not used in the example of FIG. 8, it is also possible to create a post-conversion character string “Let's run” using this rule.

  In the above embodiment, the rules and conversion processing used for the conversion have been described using the verb as an example. Similarly, the conversion rules for the adjective verbs and adjectives are converted for each utilization pattern corresponding to each part of speech. It is possible to create dictionary data for automatic reply learning by setting.

Here, a schematic procedure of processing executed by the automatic reply learning processing unit 104 will be described with reference to the flowchart of FIG.
In this process, first, the automatic reply learning dictionary 112 is initialized (step S1), and the morpheme analysis unit 105 is caused to execute morpheme analysis on the received mail to be processed (step S2).

When a predetermined number of clauses are extracted from the received mail by morphological analysis, thereafter, steps S3 to S7 are executed while paying attention to each clause until there is no unprocessed clause. Specifically, in this loop, the first clause in the unprocessed clause is extracted, the character string of this clause is set as a conversion target character string (step S4), and dictionary data is used using this conversion target character string. Is created (step S5). Further, a process of registering the created dictionary data in the reply automatic learning dictionary 112 is executed (step S6).
Finally, the phrase of the conversion target character string is set as processed (step S7), and the process returns to step S3.

When the above processing is executed for all the phrases (“NO” in step S3) and dictionary data is stored in the automatic reply learning dictionary 112 (“YES” in step S8), the automatic reply learning processing unit 104. It is effectively set the reply automatic learning dictionary 112 (step S9). After this effective setting process, conversion candidates are obtained by a search process for both the predictive conversion dictionary 111 and the automatic reply learning dictionary 112 in accordance with the assembly of the read character string on the reply mail creation screen. Will be extracted. When the creation of the reply mail is completed, the reply automatic learning dictionary 112 is invalidated.

If no dictionary data is stored in the automatic reply learning dictionary 112 (“NO” in step S8), the process ends without setting the automatic automatic reply dictionary 112 valid. In this case, in the reply mail creation process, conversion candidates are extracted by a search process for only the prediction conversion dictionary 111.
As an example in which dictionary data is not stored, a case where text data is not included in the received mail (for example, mail with an image attached) can be considered.

Next, details of the dictionary data generation process in step S5 will be described with reference to FIG.
In this embodiment, a file for storing a dictionary data list is set in the working memory 13, and in step S11, processing for initializing this dictionary data list (processing for clearing the list generated in the previous processing) is performed. Execute. Next, the part-of-speech information of the independent word included in the conversion target character string is confirmed (step S12). If the independent word is a useful word ("YES" in step S13), a loop of steps S14 to S18 is executed.

In step S14, the wording replacement table 114 is searched for the conversion target character string.
In this search, information on the utilization form of each replacement rule and an attachment word pattern in the table are collated based on the utilization pattern and utilization form of the utilization word of the conversion target character string and the attached word. As a result, when a replacement rule that matches the conversion target character string is found (“YES” in step S15), processing for converting the conversion target character string is executed using the replacement rule (step S16).

The detailed procedure for step S16 is shown in FIG. 12, and will be described below with reference to FIG.
In this conversion processing, the combination of the part-of-speech information of the character string to be converted and the end of the stem of the usage word and the usage form of the replacement pattern in the replacement rule found by the search in step S14 is set as the search key. (Step S21). Then, the utilization form setting rule that matches the search key is read from the utilization form table 115 (step S22).

  In the next step S23, the read utilization type setting rule is applied to the utilization word of the conversion target character string, and a character string representing the utilization form corresponding to the definition of the replacement rule and its reading are derived. Further, it is checked whether or not an attached word is defined in the replacement pattern in the replacement rule. If it is defined (“YES” in step S24), the attached form is attached to the utilization form derived in step S23. A combination of words is set as a converted character string (step S25). Further, the reading of the utilization form plus the reading of the attached word is set as the reading of the converted character string (step S26).

  On the other hand, if no ancillary word is defined in the replacement pattern (“NO” in step S24), only the utilization form derived in step S23 is set as the converted character string (step S27), and its utilization is performed. The shape reading is set to the converted character string reading (step S28).

  The description will be continued with reference back to FIG. By executing step S16 including the processing of steps S21 to S28, the conversion target character string is converted into a post-conversion character string having a notation that conforms to the definition of the replacement rule. In step S17, the priority is calculated by an operation of adding the standard priority and the usage frequency included in the replacement rule used for the conversion process. In step S18, a combination of the converted character string, its reading, priority, and the ID number of the replacement rule used in the conversion process is stored in the dictionary data list as dictionary data.

Thereafter, the process returns to step S14, and a process for searching for a replacement rule that matches the conversion target character string is executed. If a suitable replacement rule is found, a conversion process (step S16) and a priority calculation process (step S17) using the replacement rule are executed, and further, dictionary data based on these results is generated and stored in the dictionary list. The storing process (step S18) is executed. In the wording replacement table 114 , since a plurality of replacement rules are set for each combination of the utilization pattern and utilization form, until the replacement rule cannot be extracted from the above loop (step S15 becomes “NO”). By executing, a plurality of dictionary data can be stored in the dictionary data list.

  If the independent word of the character string to be converted is not a use word (step S13 is “NO”), the independent word is read and stored in the dictionary data list (step S19). This makes it possible to set non-use words such as nouns in dictionary data based on the words.

Next, a process for extracting dictionary data created by the processes shown in FIGS. 10 to 12 as conversion candidates will be described.
As described with reference to FIG. 11, the standard priority set in the replacement rule used for conversion and the usage frequency are added to the dictionary data derived from the conversion target character string including the use word. The selected value is set as the priority. When the automatic reply learning dictionary 112 in which the dictionary data is stored is set to be valid, the candidate extracting unit 103 uses the predictive conversion dictionary each time a reading character string is assembled in response to input of kana characters constituting the reading. 111 and the automatic reply learning dictionary 112 are searched. At this time, the candidate extraction unit 103 assigns a predetermined amount of weight to the priority of the conversion candidates extracted from the automatic reply learning dictionary 112. Is added. As a result, the conversion candidates extracted from the automatic reply learning dictionary 112 can be displayed with priority over the conversion candidates extracted from the predictive conversion dictionary 111.

For example, on the reply mail creation screen shown in FIG. 3, all six converted character strings generated by the conversion process shown in FIG. 8 are displayed. In particular, “run” derived from replacement rule 1 and “do not run” derived from replacement rule 2 each have a higher standard priority and frequency of use than others, and the above weights Is added to the first place and the second place.
By such candidate display control, the convenience when the user inputs an expression related to the utilization word in the received mail is improved, and the character string to be input can be determined efficiently.

  Furthermore, in this embodiment, when the conversion candidate extracted from the automatic reply learning dictionary 112 is selected and confirmed, the use frequency of the replacement rule created in the character string of the conversion candidate is updated. Yes. This update process is performed by the automatic reply learning processing unit 104 that has received notification of the conversion candidate determined from the control processing unit 101.

  When dictionary data is created again using the same replacement rule in the subsequent automatic learning process by updating the frequency of use of the replacement rule used to create the conversion candidate selected as described above, A higher priority than the previous time can be set for the dictionary data. Therefore, it is possible to set a state in which conversion candidates based on frequently used replacement rules are likely to be set higher.

  However, in the above update process, the selection effect cannot be applied to words having a utilization pattern different from the utilization word of the selected conversion candidate. The embodiment described next solves this problem so that a conversion candidate representing an expression of an idea similar to a conversion candidate selected in the past is preferentially displayed regardless of a difference in utilization pattern. With the goal.

  In this embodiment, in order to achieve the above object, the word replacement table 114 is configured as shown in FIG. In the replacement rule of the wording replacement table 114, a definition similar to that shown in FIG. 5 is set, and a similar ID and a conflict ID are registered. The similar ID is an ID number of a rule having a similar concept of the replacement pattern among other replacement rules, and the conflict ID is an ID number of a rule having an opposite concept of the replacement pattern among other replacement rules.

According to the example of FIG. 13, between the replacement rules 1, 2, 3, 101, 102, and 103 in which polite expressions (“mas”, “no”, “mashou”) are set as the appendix in the replacement pattern. Then, the respective ID numbers are registered as similar IDs. On the other hand, the replacement rules 7 and 107 have a similar concept to the replacement patterns of the replacement rules 2 and 102 (negative expression), but a replacement pattern of an expression that is not polite is set. This ID number is registered as a conflict ID in the replacement rules 2 and 102. Similarly, in the replacement rules 7 and 107, the ID numbers of the replacement rules 2 and 102 are registered as conflict IDs.
Note that the manner of setting the similar ID and the conflict ID is not limited to the example of FIG.

When dictionary data for automatic reply learning is created using the phrase replacement table 114 having the configuration shown in FIG. 13 and the dictionary data is displayed as a conversion candidate and selected, the automatic reply learning processing unit 104 uses the frequency of use. As the process for updating the process, the process shown in FIG. 14 is executed.

A specific procedure will be described with reference to FIG.
First, in the first step S51, the ID number of the replacement rule used for creating this conversion candidate is acquired based on the communication from the control processing unit 101 that has performed the process of determining the selected conversion candidate. This ID number can be acquired because the ID number (see step S18 in FIG. 11) included in the dictionary data when the dictionary data is created is taken over by the conversion candidate.

  When the ID number is acquired, the automatic reply learning processing unit 104 executes a process of incrementing the usage frequency for the ID number replacement rule (step S52). Further, when a similar ID is set in this replacement rule (“YES” in step S53), the use frequency of the replacement rule corresponding to the similar ID is incremented (step S54). If a conflict ID is set for the replacement rule (“YES” in step S54), the use frequency of the replacement rule corresponding to the conflict ID is decremented (step S55).

  If a plurality of similar IDs or conflicting IDs are set, the use frequency is incremented or decremented for each replacement rule corresponding to each ID in steps S53 and S55.

  According to the above process, when displaying the dictionary data created by the automatic reply learning process for the utilization word as a conversion candidate, whether or not the replacement rule used for creating the conversion candidate has been used in the past Regardless, it is possible to preferentially display conversion candidates that represent an idea similar to the expression that the user likes and selects. Therefore, conversion candidates that match the user's preferred expression are displayed at the top regardless of the type of utilization pattern, so that candidate selection operations are facilitated and operability can be improved.

  In the examples of FIGS. 13 and 14, the frequency of use of a replacement rule whose idea is similar to the replacement rule used to create the selected conversion candidate is increased, and the frequency of use of a rule whose idea is contradictory is reduced. However, the latter process is not always necessary. In addition, the unit for updating the usage frequency is not limited to one time, and a plurality of frequencies may be updated as one unit.

DESCRIPTION OF SYMBOLS 1 Mobile phone 2 Display part 3 Operation part 10 Control part 11 CPU
12 main memory 13 a working memory 20 candidate display column 31 character key 100 the character input system 101 control processor 102 reads the character string assembly unit 103 candidate extraction unit 104 replies the automatic learning processing unit 105 the morphological analysis unit 112 replies dictionary 113 morpheme for automatic learning Dictionary for analysis 114 Phrase replacement table 115 Practical table 200 Mailer

Claims (5)

A program for causing a computer having a function of transmitting and receiving an e-mail to execute a process of inputting a character string corresponding to the operation to a mail to be transmitted according to an operation of inputting characters,
For multiple types of usage patterns, as a conversion rule for replacing the expression in the usage form of the pattern with another expression, the combination of the usage form and the adjunct after replacement with the combination of the usage form before the replacement and the adjunct Conversion rule storage means for storing a conversion rule having a configuration associated with
A post-conversion character string storage means for storing a post-conversion character string created by applying the conversion rule;
Morphological analysis means for performing morphological analysis on one of the received emails;
In response to an operation for instructing a reply to one of the received mails, the morpheme analysis unit includes morpheme analysis for the reply target mail , and then includes a use word extracted by the morpheme analysis. Using the phrase as a processing target for each phrase, the conversion rule storage means is searched by the combination of the usage pattern of the word used in the character string of the processing target clause and the usage pattern and the attached word in the processing target clause, and the combination is obtained. A step of extracting a suitable conversion rule, and a step of creating a converted character string by applying a combination of a replacement usage form and an attached word set in the extracted conversion rule to a processing target clause. performed on the processing target clause, while the reply mail creation screen is raised with respect to the operation of the post-conversion string produced instructs the reply Ri as enabled, the learning processing means for storing the converted character string storage means in association with reading of the converted character string,
When the operation to input the reading of the character string to be input is performed on the reply e-mail creation screen launched in response to the operation to instruct the reply, the converted character string storage unit is in an effective state Candidate extraction means for performing a search using the input reading on the converted character string stored in the above, and extracting the converted character string matching the reading as a conversion candidate,
Candidate display means for displaying a candidate display field containing conversion candidates extracted by the candidate extraction means on the reply mail creation screen;
An input character string determining means for determining a character string of the selected conversion candidate as an input character string to the reply mail in response to an operation of selecting one of the conversion candidates displayed in the candidate display field;
A program for inputting characters, characterized by causing the computer to function as each means. The learning processing means executes processing for each processing target clause after clearing the converted character string storage means in response to an operation to instruct a reply to one of the received mails,
The program for character input according to claim 1, wherein the candidate extraction unit executes a search for the converted character string storage unit after being cleared in response to the operation to instruct the reply . In the program according to claim 1 or 2,
Each conversion rule stored by the conversion rule storage means includes information indicating the frequency of use of the conversion rule, and the converted character string displayed as a conversion candidate by the candidate display means is determined as the input character string. A program that causes the computer to function as a use frequency update unit that updates the use frequency of the conversion rule used for conversion to the confirmed character string when the input character string is confirmed by the means as the input character string to the reply mail; Included,
The candidate display means, when a plurality of converted character strings are extracted by the candidate extraction means, the display order of these converted character strings, the frequency of use of the conversion rule used for conversion to each character string A program for character input that is determined using. In the program according to claim 3,
Each conversion rule stored in the conversion rule storage means is associated with each of the conversion concepts that are similar in concept.
The usage frequency update unit updates the usage frequency of the conversion rule associated with the conversion rule together with the usage frequency of the conversion rule used for conversion into the character string determined by the input character string determination unit. A program for character input. In a communication terminal device comprising a transmission / reception unit for transmitting and receiving an e-mail, an operation unit including a plurality of keys assigned with kana characters, and a display unit for displaying an e-mail document creation screen,
For multiple types of usage patterns, as a conversion rule for replacing the expression in the usage form of the pattern with another expression, the combination of the usage form and the adjunct after replacement with the combination of the usage form before the replacement and the adjunct Conversion rule storage means for storing a conversion rule having a configuration associated with
A post-conversion character string storage means for storing a post-conversion character string created by applying the conversion rule;
Morphological analysis means for performing morphological analysis on one of the received emails;
In response to an operation for instructing a reply to one of the received mails, the morpheme analysis unit includes morpheme analysis for the reply target mail , and then includes a use word extracted by the morpheme analysis. Using the phrase as a processing target for each phrase, the conversion rule storage means is searched by the combination of the usage pattern of the word used in the character string of the processing target clause and the usage pattern and the attached word in the processing target clause, and the combination is obtained. A step of extracting a suitable conversion rule, and a step of creating a converted character string by applying a combination of a replacement usage form and an attached word set in the extracted conversion rule to a processing target clause. performed on the processing target clause, while the reply mail creation screen is raised with respect to the operation of the post-conversion string produced instructs the reply Ri as enabled, the learning processing means for storing the converted character string storage means in association with reading of the converted character string,
When the operation to input the reading of the character string to be input is performed on the reply e-mail creation screen launched in response to the operation to instruct the reply, the converted character string storage unit is in an effective state Candidate extraction means for performing a search using the input reading on the converted character string stored in the above, and extracting the converted character string matching the reading as a conversion candidate,
Candidate display means for displaying a candidate display field containing conversion candidates extracted by the candidate extraction means on the reply mail creation screen;
An input character string determining means for determining a character string of the selected conversion candidate as an input character string to the reply mail in response to an operation of selecting one of the conversion candidates displayed in the candidate display field;
A communication terminal device comprising the means described above.

Images