JP2012068857A - Information processor, email reply sentence extraction method and program therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enhance convenience of a mail function by setting an optimum searching range for a received mail text to improve extraction accuracy of reply sentence candidates.SOLUTION: An information processor comprising an email creation function divides a received mail text into a plurality of substrings by a division function unit (8). For one or all of a plurality of consecutive substrings, the information processor selects a received sentence candidate identical with or most similar to the substring(s) as a received sentence by referring to a storage unit (12). The storage unit (12) stores the received sentence candidate corresponding to one or a plurality of consecutive substrings and reply sentence information associated with the received sentence candidate. Then, the information processor sets a combination of received sentences which makes all divided substrings to become continuous substrings and is most similar to the mail text as a searching range, and based on the searching range, extracts the reply sentence information corresponding to the received sentence stored in the storage unit (12).

Description

The present invention relates to an information processing apparatus that transmits and receives an electronic mail, and extracts and displays a reply sentence candidate stored in a database or the like according to the content of the received electronic mail when creating a reply sentence. The present invention relates to an e-mail reply sentence extraction method and a program thereof.

  Communication using an electronic mail function installed in a cellular phone or the like is indispensable, and many users think that they should reply immediately to received mail. Also, the user side who sends the mail often expects a quick reply from the receiver side.

  In order to speed up the response to received mail, it may be possible to easily create mail. For example, a reply mail document is automatically prepared according to the contents of the received mail, and the user can Some reduce the time and effort of inputting a document.

  In order to automatically create and send such a reply email, register a keyword and reply message in advance, and if the received email contains a keyword, reply the registered message, Some transfer to a registered transfer destination (for example, Patent Document 1). Further, past received mails and their reply mails are stored in association with each other, and when a new mail arrives, a similar past mail document is searched as a search key. If it is determined that the similarity is high, all or part of the searched email will be automatically replied, or the respondent will add a comment depending on the similarity of the searched email. (For example, patent document 2).

  When an email containing a user name or inquiry keyword is received, the inquiry content is classified based on the user identification or keyword with reference to the user classification table, and related cases are collectively searched based on the classification result. There is one that automatically replies an answer (for example, Patent Document 3). In addition, regarding the information provided by product catalogs, WWW, etc., the received question content is automatically decoded, the corresponding answer is automatically returned from the accumulated answers, and it is suitable for the accumulated answers When there is no answer, there is one that notifies the respondent and accumulates a new answer (for example, Patent Document 4).

The answer information for the inquiry is accumulated, and the keyword is searched based on the analysis result of the inquiry content to find the answer information. Then, the answer accuracy of the answer information is measured, and if it exceeds a predetermined value, an answer is automatically created, and if it is less than the predetermined value, there is a type that prompts an input of whether or not the answer is correct (for example, patent Reference 5). Also, feature information is extracted from the syntax analysis result for the received mail, and a reply mail is generated by using the reply template in contrast to the feature information. In addition, when there are multiple sentence examples for feature information, it also has a function to accept which one to select. For feature information that is a question sentence, an affirmative sentence and a negative sentence for the question sentence (For example, Patent Document 6).

JP-A-2-241241 Japanese Patent Laid-Open No. 11-275137 JP 2000-20541 A JP 2000-57066 A Japanese Patent Laid-Open No. 2002-132661 JP 2002-49564 A

  By the way, in a cellular phone or the like, due to downsizing of the casing, for example, a plurality of characters are assigned to one operation button, or it is difficult to press down due to downsizing of the operation button. It is difficult to create a sentence quickly.

  In addition, when presenting previously stored reply text, etc. for the received mail text, for example, when the text of the mail text is created in colloquial style, etc., an appropriate response can be made depending on how to refer to the reply text candidate. There are cases where no candidate is listed. For example, even if a word or the like partially matches the mail text, there is a possibility that the content of the entire document may be captured in a different meaning.

  Regarding such demands and problems, Patent Documents 1 to 6 do not disclose or suggest them, and do not disclose or suggest a configuration or the like for solving them.

  Therefore, the purpose of the information processing apparatus, the e-mail reply sentence extraction method, or the program thereof of the present disclosure is to extract a stored reply sentence candidate by optimizing the search range setting process for the received mail body. The purpose is to increase accuracy.

Another object of the information processing apparatus, e-mail reply sentence extraction method, or program thereof according to the present disclosure is to improve the convenience of the mail function by presenting appropriate reply sentence candidates to the received mail text. There is.

  In order to achieve the above object, an information processing apparatus according to the present disclosure includes a division function unit, a storage unit, a received text selection unit, a search range setting unit, and a reply text extraction unit. The division function unit divides the received mail body into a plurality of partial character strings. The storage unit stores received sentence candidates corresponding to one or a plurality of consecutive partial character strings and reply sentence information associated with the received sentence candidates. The received sentence selection unit selects, as the received sentence, the received sentence candidate that matches or is most similar to each of one or a plurality of consecutive partial character strings. The search range setting unit sets, as a search range, a combination of the received texts in which all of the divided partial character strings are continuous and most similar to the mail text. The reply sentence extraction unit extracts reply sentence information corresponding to the received sentence stored in the storage unit based on the search range. With this configuration, the above object is achieved.

  In order to achieve the above object, an e-mail reply sentence extraction method of the present disclosure is an e-mail reply sentence extraction method of an information processing apparatus, which includes a division process, a storage process, a received sentence selection process, and a search range. A setting process and a reply sentence information extraction process are included. The dividing process divides the received mail body into a plurality of partial character strings. In the storage process, a received sentence candidate corresponding to one or a plurality of consecutive partial character strings and reply sentence information associated with the received sentence candidate are stored in the storage unit. In the received sentence selection process, the received sentence candidate that matches or is most similar to all of one or a plurality of consecutive partial character strings is selected as a received sentence. In the search range setting process, all of the divided partial character strings are continuous, and the combination of the received text that is most similar to the mail body is set as the search range. The reply text information extraction process extracts reply text information corresponding to the received text stored in the storage unit based on the search range. With this configuration, the above object is achieved.

In order to achieve the above object, an e-mail reply sentence extraction program of the present disclosure is an e-mail reply sentence extraction program to be executed by a computer, wherein the computer has a division function, a storage function, and a received sentence selection function, The search range setting function and the reply sentence information extraction function are executed. The division function divides the received mail body into a plurality of partial character strings. The storage function causes the storage unit to store received sentence candidates corresponding to one or a plurality of consecutive partial character strings and reply sentence information associated with the received sentence candidates. The received sentence selection function selects the received sentence candidate that is the same or most similar as the received sentence for all of one or a plurality of consecutive partial character strings. The search range setting function sets a combination of the received texts that is the most similar to the mail text as a search range in which all the divided partial character strings are continuous. The reply text information extraction function extracts reply text information corresponding to the received text stored in the storage unit based on the search range. With this configuration, the above object is achieved.

  According to the information processing apparatus, e-mail reply sentence extraction method, or program thereof of the present disclosure, any of the following effects can be obtained.

  (1) By setting the search range so that the mail body and the stored received sentence candidate are the closest to each other, it is possible to extract appropriate reply sentence candidates and send a response mail that matches the contents of the mail body. Can be created.

  (2) By extracting appropriate reply sentence candidates according to the contents of the mail body, the user can create a response mail with only a selection operation, reducing the labor of character input operation, and prompt response mail Can be made.

(3) Since the reply sentence candidate for the content is extracted and displayed along the text described in the mail text, it is possible to respond to the written contents of the sender of the mail without omission.

Other objects, features, and advantages of the present invention will become clearer with reference to the accompanying drawings and each embodiment.

It is a figure which shows the function structural example of the information processing apparatus which concerns on 1st Embodiment. It is a figure which shows an example of the extraction principle of the combination of the received sentence candidate most similar with respect to a received mail. It is a figure which shows an example of the principle which extracts a reply sentence candidate based on the set search range. It is a flowchart which shows an example of a reply mail preparation process. It is a figure which shows the function structural example of the information processing apparatus which concerns on 2nd Embodiment. It is a figure which shows the external appearance structural example of a mobile telephone. It is a figure which shows the hardware structural example of a mobile telephone. It is a figure which shows an example of the division | segmentation process of a received mail. It is a figure which shows an example of the extraction process of a received sentence candidate. It is a figure which shows an example of a received sentence / reply sentence database. It is a figure which shows the specific example of a received sentence extraction process. It is a figure which shows an example of the combination extraction principle of the partial character string continuous by similarity calculation. It is a figure which shows an example of the calculation principle of the partial character string continuous similarity total. It is a figure which shows an example of the extraction principle of a prior | preceding partial character string using the partial similarity accumulation of a partial character string continuous. It is a figure which shows an example of the extraction principle of a prior | preceding partial character string using the partial similarity accumulation of a partial character string continuous. It is a figure which shows an example of the extraction principle of a prior | preceding partial character string using the partial similarity accumulation of a partial character string continuous. It is a figure which shows an example of the determination process of the partial character string continuous combination. It is a figure which shows an example of the extraction process of a reply sentence candidate. It is a figure which shows an example of the state as which the reply sentence was selected. It is a figure which shows an example of the display state of a received mail. It is a figure which shows an example of the display state of the division | segmentation received mail and the extracted reply sentence candidate. It is a figure which shows the example of a display of the preparation screen of a reply mail. It is a flowchart which shows an example of a reply mail preparation process. It is a flowchart which shows an example of a database search process. It is a flowchart which shows an example of the combination determination process of a partial character string continuous. It is a flowchart which shows an example of the combination determination process of a partial character string continuous. It is a figure which shows the reply mail preparation function structural example which concerns on 3rd Embodiment. It is a figure which shows the structural example of a received sentence / reply sentence database. It is a flowchart which shows an example of a reply mail preparation process. It is a figure which shows the example of a reply mail preparation function structure based on 4th Embodiment. It is a figure which shows the example of a display of the produced reply mail. It is a flowchart which shows an example of a reply mail preparation process. It is a figure which shows the reply mail preparation function structural example which concerns on 5th Embodiment. It is a figure which shows the structural example of a received sentence / reply sentence database. It is a figure which shows an example of extraction and replacement processing of specific information. It is a flowchart which shows the example of a reply mail preparation process. It is a flowchart which shows the example of a reply mail preparation process. It is a figure which shows the reply mail preparation function structural example which concerns on 6th Embodiment. It is a figure which shows the structural example of an expression replacement character string table. It is a figure which shows an example of an expression substitution process and a database search process. It is a flowchart which shows an example of a reply mail preparation process. It is a figure which shows the reply mail preparation function structural example which concerns on 7th Embodiment. It is a figure which shows the structural example of a received sentence / reply sentence database. It is a flowchart which shows an example of a reply mail preparation process. It is a figure which shows the external appearance structural example of PDA which concerns on other embodiment.

[First Embodiment]

  The first embodiment will be described with reference to FIG. 1, FIG. 2, and FIG. FIG. 1 is a diagram illustrating an example of a mail creation function configuration according to the first embodiment, FIG. 2 is a diagram illustrating an example of an extraction principle of a combination of received sentence candidates most similar to a received mail, and FIG. FIG. 10 is a diagram illustrating an example of a principle of extracting reply sentence candidates based on a set search range. The configuration illustrated in FIGS. 1 to 3 is an example, and the present invention is not limited to this.

  The information processing apparatus 2 is an example of the information processing apparatus of the present disclosure, and in creating a reply mail for a received electronic mail, a mail creation apparatus 4 that provides a reply sentence candidate corresponding to the description content of the text of the electronic mail Is configured. In this mail creation device 4, the received sentence that matches or is most similar to the contents of the received mail 6 is extracted from a plurality of received sentence candidates stored in the database. Then, a reply sentence candidate associated with the extracted received sentence is presented. In order to extract a received sentence corresponding to the entire contents of the received mail 6, the mail creating apparatus 4 sets an optimum search range for extracting a received sentence candidate.

  Therefore, the mail creating apparatus 4 having functions such as setting a search range for the mail text, extracting an optimum received sentence candidate, and extracting a reply sentence candidate divides the received mail 6 into partial character strings by the dividing function unit 8. . Then, the received sentence selection unit 10 selects a received sentence that matches or is similar to the received sentence candidates in the storage unit 12 for one or a plurality of consecutive partial character strings. The search range setting unit 14 sets a combination of one or a plurality of partial character strings that match or are similar to the entire mail body as a search range, and the reply sentence extraction unit 16 extracts reply sentence candidates. I do.

  The division function unit 8 is an example of a unit that takes the received mail 6 and divides the mail body into partial character strings configured for each predetermined unit. The mail text may be based on, for example, punctuation marks, pictograms, emoticons, etc. included in the mail text, or divided into partial character strings in units of a predetermined number of characters or length, such as for each word or each phrase. . For example, as shown in FIG. 2A, for example, as shown in FIG. 2B, the mail division processing is performed on a mail text composed of a character string of a predetermined length as shown in FIG. The character strings are divided into partial character strings S0 to S3. Here, the mail text shown in A of FIG. 2 shows, for example, the case of one paragraph having a predetermined length, but is not limited thereto. For example, a division process may be performed in a similar manner even for an email composed of a plurality of paragraphs with paragraphs lowered for each short sentence based on the contents of the email, or a partial character string may be set for each paragraph.

  The received sentence selection unit 10 is an example of a search function that selects and extracts the most similar received sentence from the received sentence candidates stored in the storage unit 12 for one or a plurality of consecutive partial character strings. As shown in FIG. 2C, the received sentence selection unit 10, for the divided partial character strings S <b> 0 to S <b> 3, has a partial character string sequence of various lengths composed of a single partial character string or a continuous partial character string. Is set as the search range, and the most similar received sentence is searched from the database. As a result of this search, for example, the received text in each search range is stored in the storage unit 12. In this case, for example, a calculated value, similarity information, or the like that is determined whether or not each search is similar may be stored together.

  The memory | storage part 12 is an example of the memory | storage means in which the received sentence and one or more reply sentence information matched with it were memorize | stored, Comprising: For example, it is comprised with the database. The storage unit 12 stores received sentence candidates having various contents corresponding to one or a plurality of continuous partial character strings. The received sentence candidate stored in the storage unit 12 is a character string in which one reply sentence is expected with respect to the content of one or a plurality of consecutive partial character strings, and is limited to the case where it is composed of one sentence. Instead, it may be composed of a plurality of sentences. Moreover, you may be comprised by 1 phrases, such as a clause.

  The search range setting unit 14 is an example of a unit that sets, as a search range, a combination of received sentences in which all divided partial character strings are continuous and most similar to the content of the mail text. In this search range setting, for example, as shown in FIG. 2C, a combination of partial character strings similar to the entire mail body is determined based on search results for consecutive partial character strings of various lengths. In the partial character string combination determination process, the similarity between consecutive partial character strings is referred to, and a combination in which the divided partial character strings are continuous is selected for the mail body. As shown in FIG. 3D, for the selected combination, the partial character string continuation is determined so that the partial character strings S0 to S3 are continuous. Based on the set search range, the most similar received sentence is extracted for each successive partial character string and set as the search result of this mail.

  The reply sentence extraction unit 16 is an example of a means for extracting reply sentence information corresponding to the received sentence stored in the storage unit 12 based on the set search range. As shown in E of FIG. The reply sentence candidate is presented to the user. When there are a plurality of reply sentence candidates, the list is presented to prompt the user to select one sentence. The reply sentence candidates to be presented may be presented in the order stored in the storage unit 12. Further, for example, the presentation order and the number of presentations may be changed according to the usage frequency, the sender of the mail text, or the like.

  Next, FIG. 4 is referred with respect to a reply mail creation process. FIG. 4 is a flowchart illustrating an example of a reply mail creation process. The processing content, processing procedure, and the like shown in FIG. 4 are examples and are not limited to these.

  This reply e-mail creation process is an example of an e-mail reply sentence extraction method of the present disclosure and a program thereof, and includes a process of setting a search range for extracting a reply sentence having an optimum content for a received mail. .

  Therefore, in the reply mail creation process, the received mail 6 is divided into partial character strings for each predetermined unit (step S11). Then, for the divided partial character strings, the received sentence candidates in the storage unit 12 are searched for each successive partial character string obtained by combining one or a plurality of consecutive character strings, and the most similar received sentence is selected (step S12).

  For the received text selected for partial character string continuations of various lengths, a combination of similar partial character string continuations is determined for the entire mail body, and this combination is set as a search range for the storage unit 12 (step S13). Based on the set search range, reply sentence candidates are extracted from the storage unit 12 and displayed (step S14).

  According to such a configuration, an appropriate reply sentence candidate can be extracted by setting the search range so that the mail text and the stored received sentence candidate are closest to each other, and according to the contents of the mail text Response mail can be created. In addition, by extracting appropriate reply sentence candidates according to the content of the mail text, the user can create a response mail only with the selection operation. Can be made. In addition, since reply sentence candidates for the contents are extracted and displayed along the text described in the mail text, it is possible to respond to the contents described by the sender of the mail without omission.

[Second Embodiment]

  Next, FIG. 5 is referred about 2nd Embodiment. FIG. 5 is a diagram illustrating a functional configuration example of the information processing apparatus according to the second embodiment. The configuration shown in FIG. 5 is an example, and the present invention is not limited to this. In FIG. 5, the same components as those in FIG. 1 are denoted by the same reference numerals, and the description thereof is omitted.

  This information processing device 2 is an example of the information processing device of the present disclosure, and includes the above-described mail creation device 4, and sets an optimal search range for the received sentence candidate stored in the storage unit. The reply sentence candidates extracted based on the search range are presented to the user. And a reply mail with respect to the received mail 6 can be automatically created by selecting a specific reply sentence from the presented reply sentence candidates for all search ranges.

  Therefore, this mail creation device 4 includes, for example, the search range setting function and the like by the reply sentence extraction part 16 as well as the mail division part 20, the search part 22, the received sentence / reply sentence database 24, and the reply sentence combining part 26. A reply sentence extraction function or the like can be realized. The information processing apparatus 2 includes a reply sentence candidate display section 28, a selection input section 30, and the like together with the mail creation apparatus 4.

  The mail dividing unit 20 is an example of means for dividing the mail body into partial character strings for each predetermined unit, and constitutes the above-described dividing function unit 8. The character string included in the received mail 6 is divided on the basis of, for example, each punctuation mark, each word, or each phrase.

  The search unit 22 is an example of a unit that searches a similar received sentence from a database with various lengths from a single case to a continuation of a plurality of partial character strings with respect to the divided partial character strings. The search unit 22 includes, for example, the received text selection unit 10 and the search range setting unit 14 described above, and sets an optimal search range for the received text / reply text database 24. The received sentence selection unit 10 includes, for example, a similarity calculation unit 23 that calculates a similarity between a partial character string obtained by dividing the received mail 6 and a character string included in a received sentence candidate in the received sentence / reply sentence database 24. I have.

  The received sentence / reply sentence database 24 stores received sentence candidates to be searched for one or a plurality of divided partial character strings and reply sentence candidates associated with the received sentence candidates. The storage unit 12 is configured.

  The reply sentence combining unit 26 is an example of a means for creating a reply mail by connecting the reply sentences selected by the user to the presented reply sentence candidates.

  The reply sentence candidate display unit 28 is an example of a unit that presents a list of reply sentence candidates extracted on the display screen of the information processing apparatus 2 and prompts the user to select. Presentation of the reply sentence candidate is not limited to using the display screen, and for example, presentation by voice may be performed.

  The selection input unit 30 is an example of means for performing a decision input operation for a reply sentence candidate selected by the user on the presented list of reply sentence candidates, and includes, for example, a cursor key and a decision key of the information processing apparatus 2 Or you may comprise with the touchscreen display means which comprised the operation part and the display part integrally. Then, based on the result of the selection operation, the reply message combining unit 26 creates a reply mail 31.

  Next, a configuration example of the information processing apparatus will be described with reference to FIGS. FIG. 6 is a diagram illustrating an external configuration example of a mobile phone, and FIG. 7 is a diagram illustrating a hardware configuration example of the mobile phone. The configuration shown in FIGS. 6 and 7 is an example, and the present invention is not limited to this.

  The mobile phone 40 is an example of an information processing apparatus according to the present disclosure, and constitutes an electronic mail apparatus having the electronic mail transmission / reception function and the reply mail creation function described above. In the cellular phone 40, for example, an operation-side housing unit 42 and a display-side housing unit 44 are connected by a hinge unit 46 so as to be opened and closed.

  For example, the operation-side housing unit 42 includes an operation input unit 48 and includes a keyboard to which cursor keys, determination keys, letters, numbers, and the like are assigned, and functions as the selection input unit 30 described above. Good.

  The display-side housing unit 44 includes, for example, a display unit 50 configured by an LCD (Liquid Crystal Display) or the like, and may function as a received mail display function or the above-described reply sentence candidate display unit 28.

  Although the mobile phone 40 is shown as an example of the information processing apparatus of the present disclosure, the configuration is not limited to the configuration illustrated in FIG. 6. For example, the operation-side casing unit 42 and the display-side casing unit 44 are integrally configured. It may be configured so that it can be slid and stacked on top and bottom. Moreover, the structure which the operation side housing | casing part 42 and the display side housing | casing part 44 isolate | separated may be sufficient.

  As shown in FIG. 7, the hardware configuration of the cellular phone 40 includes, for example, a processor 52, a wireless communication unit 54, a RAM (Random Access Memory) 56, a storage unit 58, a display control unit 66, a voice input / output unit. 68 and the operation input unit 48 or the like.

  The processor 52 is an example of a unit that performs arithmetic processing of a control program such as an OS (Operating System) and a mail creation program stored in the storage unit 58, and is an example of a control unit of the mobile phone 40 that is an information processing apparatus. It functions as a part. The processor 52 may be configured by a CPU (Central Processing Unit), for example. Then, as described above, it is possible to execute mail division processing, similar received sentence extraction processing, appropriate search range setting processing, reply sentence candidate extraction processing, and the like by executing operations such as a mail creation program.

  The wireless communication unit 54 is a communication unit that performs voice communication and data communication such as a Web (World Wide Web) page in addition to an electronic mail function, and includes an antenna 55 to perform wireless communication with a base station. That is, by providing the wireless communication unit 54, the cellular phone 40 can execute a telephone function and a communication function.

  The RAM 56 constitutes a work area for executing a mail creation program stored in the storage unit 58, and the processor 52 performs arithmetic processing of the control program, whereby the mail division unit 20 and the search unit 22 described above. , Function as the reply sentence extraction unit 16, reply sentence combination unit 26, and the like.

  The storage unit 58 is an example of a storage unit or a recording unit of the mobile phone 40 and includes, for example, a program storage unit 60 for storing a control program, and a data storage unit 62 for storing acquired data and a database. ing.

  The program storage unit 60 is an example of a unit that stores a control program such as the OS of the mobile phone 40 and a mail creation program. For example, the program storage unit 60 may be configured by a ROM (Read Only Memory).

  The data storage unit 62 stores, for example, the received sentence / reply sentence database 24. In addition, for example, received sentence candidate information extracted from the received sentence / reply sentence database 24, similarity information thereof, etc., or reply sentence information selected by the user, for each one or a plurality of consecutive partial character strings An extraction information buffer 64 or the like is provided. The data storage unit 62 may be composed of, for example, a ROM or a flash memory.

  The program storage unit 60 and the data recording unit 62 may be configured by, for example, an EEPROM (Electrically Erasable and Programmable Read Only Memory) that can electrically rewrite the contents.

  The mail creation program is not limited to those stored in the program storage unit 60, the data storage unit 62, or the like. For example, you may utilize what is memorize | stored in the recording medium which can be read by computers, such as a magnetic disk, a flexible disk, and an optical disk magneto-optical disk. Furthermore, a program or the like may be read from a server or database on the network and used.

  The operation input unit 48 is an example of a unit that detects an input operation on the mobile phone 40 and notifies the processor 52 of the detection result to perform character input or the like, and constitutes the selection input unit 30 described above. The operation input unit 48 may be configured with, for example, a keyboard and operation buttons.

  The display control unit 66 is an example of a unit that displays a list of received mails and reply sentence candidates, a selected reply sentence, a created reply mail, and the like on the display unit 50 configured by an LCD or the like. The reply sentence candidate display unit 28 is configured. In addition, for example, a so-called touch panel in which the display control unit 66 and the operation input unit 48 are integrally configured and operated by touching the display unit 50 may be used.

  The voice input / output unit 68 is an example of a means for inputting or outputting a telephone voice or the like. The voice signal is input through the microphone 70, and the voice signal is output from the receiver 72 from the radio signal received by the radio communication unit 54.

  Next, the search range setting process for the received mail will be described with reference to FIGS. 8, 9, 10, 11, 12, 13, 14, 15, 16, and 17. FIG. 8 is a diagram showing an example of received mail division processing, FIG. 9 is a diagram showing an example of received text candidate extraction processing, FIG. 10 is a diagram showing an example of a received text / reply text database, and FIG. FIG. 10 is a diagram illustrating a specific example of received sentence extraction processing. FIG. 12 is a diagram illustrating an example of a partial character string continuous combination extraction principle based on similarity calculation, FIG. 13 is a diagram illustrating an example of a partial character string continuous similarity accumulation calculation principle, and FIGS. FIG. 16 is a diagram illustrating an example of the extraction principle of the preceding partial character string using the similarity accumulation of partial character strings continuous. FIG. 17 is a diagram illustrating an example of a process for determining a combination of consecutive partial character strings. The processing contents, processing procedures, and the like shown in FIGS. 8 to 17 are examples, and are not limited to these.

  First, an example of processing for extracting a received sentence candidate and a reply sentence candidate from the database based on the set search range will be described.

  (a) Incoming mail split processing

  In the dividing process for the received mail 6, as shown in FIG. 8A, for example, the received mail 6 composed of a plurality of sentences is divided into partial character strings having a length of about a sentence. In this division process, for example, the arrangement position of a punctuation mark “.”, An exclamation mark “!”, A question mark “?”, A pictograph, an emoticon, etc. is executed as a clue. These symbols and characters are often written at the end of a sentence, and are suitable, for example, for dividing a received mail into predetermined meanings. In addition, you may set the character, symbol, etc. which are described at the end of a sentence.

  Then, when division processing is performed based on such division criteria, the received mail 6 is divided into, for example, five partial character strings from No. 0 to No. 4 as shown in FIG. 8B.

  Note that the method of dividing the received mail 6 is not limited to this standard, and the received mail may be divided at any position or character by character as long as the contents of the mail body can be accurately divided.

  (b) Received sentence candidate selection process

  Next, similar received texts are searched from the received text / reply text database 24 based on the set search range. Here, as the partial character string continuation to be searched, for example, partial character strings No. 1 and No. 2 are collectively set as one search range (partial character string continuation), and partial character strings No. 0, No. 3, No. 4 are set. Indicates a case where the search range is used alone.

  As shown in FIG. 9, the received sentence / reply sentence database 24 has, for example, a received sentence column 80 storing received sentence candidates set for each sentence, each phrase, or each word, and each received sentence. The reply message candidate field 82 stores a plurality of types of associated reply sentences. In the mail creation process, in order to present a reply sentence candidate to the user, a received sentence candidate that matches the contents of the received mail for each successive partial character string set as the search range is selected and extracted.

In this received sentence candidate extraction, the similarity between the characters included in the partial character string sequence and the characters included in the received sentence candidate is calculated. This similarity may be calculated by the similarity calculation unit 23 of the search range setting unit 14. There are various calculation methods. For example, Tatsuo Yamashita et al. “Similarity calculation method and search result presentation method effective for translation support” (Procedure of the 9th Annual Conference of the Association for Natural Language Processing) You may use what is shown in. In this method, the coincidence rate is calculated as an example of the similarity. This match rate is:
Match rate = (A × 2) / (B + C) (1)
It becomes. In this formula (1), A represents the number of characters matched in two sentences, B represents the number of characters of a sentence serving as a search key, and C represents the number of characters of a sentence in the database. Note that the method of calculating the similarity is not limited to this.

  By using the similarity (matching rate) obtained by the equation (1), the most similar received sentence can be extracted for each partial character string sequence as shown in FIG.

  Here, an example of the search process of the received sentence / reply sentence database for characters included in the partial character string sequence will be described.

  The received sentence / reply sentence database 24 includes, for example, an index section 24A and a data section 24B as shown in FIG. The index section 24A includes a heading column 84 that includes a heading character to be searched, and a data number column 86 that indicates a data number that includes the heading character. Further, the data section 24B is configured with a data number column 88 representing the data number together with the received text column 80 and the reply text candidate column 82 described above. The data number presented in the data number column 86 of the index portion 24A corresponds to the data number column 88 of the data portion 24B.

  Therefore, for example, if the search target includes the characters “movie”, referring to the index portion 24A, the data including such characters is the data number “00501” or “00502” of the data portion 24B. It is shown.

  For the partial character string to be searched, for example, the index part 24A of the received sentence / reply sentence database 24 is checked for all the included characters, the data of the data number described therein is extracted, and the partial character to be searched Calculate similarity to a column. For example, as shown in FIG. 11, when the search target partial character string (S1 + S2) is “I want to watch a movie. How about tomorrow?”, The data number “00501” including the first character “movie”, “00502” is obtained from the index part 24A, the data is obtained from the data part 24B, and the similarity is calculated.

  For example, the received sentence candidate with the data number “00501” has a similarity of 32 [%], and the received sentence candidate with the data number “00502” has a similarity of 47 [%]. Similarly, for the character “picture”, the index portion 24A is similarly examined to obtain data including the character. Here, the data already calculated for the similarity is skipped without being calculated again. In this way, the index part 24A is examined for all characters of the partial character string to be searched, data including the character is obtained, and similarity calculation is performed.

  Then, in all acquired data, the data having the maximum similarity is determined as a search result. Here, in the case of the data number “00502,” the similarity is determined as a search result with 47%. The similarity data calculated here may be stored, for example, in the extraction information buffer 64 and subjected to comparison processing.

  In the example shown in FIG. 10, the case where one character is used as a heading is shown, but a character string composed of a plurality of characters may be used as a heading. If the character string of the headline is lengthened, the number of data for one headline is narrowed down, so that the speed of search is increased while the size of the index is increased.

  (c) Search range setting process

  In the received sentence candidate extraction process described above, an optimum received sentence is extracted by combining the divided partial character strings with, for example, No. 1 and No. 2 in the search range. The division into partial character strings is based on a specific symbol, and is not necessarily divided at a semantic break that requires one reply sentence. Therefore, an object is to extract a reply sentence in a more appropriate unit by searching a database with a variable-length character string combining one or more partial character strings.

  For example, “Otsukasesa-n (^^) /” in the received mail 6 and “Otsukase-sama” in the received sentence / reply sentence database 24, 1 of the database for one partial character string. There may be cases where one received sentence corresponds. Further, for example, two partial character strings “How do you want to see a movie?” In the partial character string S1 and “How tomorrow?” In the partial character string S2, “Tomorrow, Can you go to see a movie? "

  Therefore, in this search range setting process, a received sentence having the maximum similarity with respect to one or more consecutive partial character strings (partial character string continuous) is obtained from the received sentence / reply sentence database 24, and the similarity is obtained. Set a combination that maximizes the total of

  When the received mail 6 is divided into partial character strings S0 to S3, for example, as shown in A of FIG. 12, the combination of partial character string sequences Tn composed of one or more partial character strings is n = 0-9. There are 10 patterns. This n represents a counter of a combination pattern of partial character string continuation T in which partial character strings S are arranged from the head of the received mail 6.

  As shown in FIG. 12B, the database 24 is searched for each pattern of the partial character string continuation Tn from the counter n = 0 to 9, and the similarity C (Tn) for the extracted received sentence data with the maximum similarity is extracted. ) Is calculated. The extraction of the received sentence data that maximizes the similarity may be performed using the received sentence candidate extraction process described above.

  Here, in the calculation of the similarity using the coincidence ratio of the expression (1), each similarity C (Tn) is 100 [%] at the maximum. In the search processing of the partial character string continuation Tn and the database 24, for example, the received sentence can be extracted more accurately by searching with a longer character string. For example, if the database is searched based on short words or the like, the degree of similarity increases, but an appropriate received sentence is not always extracted for the content included in the received mail 6. Therefore, in the case of partial character string continuation composed of a plurality of partial character strings so that priority is given to partial character string continuation T including more partial character strings S, for example, the number of partial character strings included is multiplied. To weight.

  Then, the similarity C (Tn) is obtained for all the partial character string continuations Tn.

  For the calculated similarity to each partial character string sequence Tn, the combination of partial character string sequences T that maximizes the cumulative similarity Ctotal (Tn) is determined as the total similarity for the entire received mail 6. That is, this combination is a search range in which a combination of received sentences that is most similar to the body of the received mail 6 can be extracted. Further, in determining the combination of the partial character strings T that maximizes the similarity, it is determined whether or not the partial character string continuation T can be preceded with respect to all the partial character string continuations T. This determination is based on the condition that all the partial character strings S are continuous with the received mail 6 and that the partial character string continuous T is combined so that they are not selected redundantly.

  In order to obtain a combination of consecutive partial character strings having the maximum similarity, for example, a determination method using a Viterbi algorithm may be used.

  In the calculation of the total value Ctotal (Tn) of similarity, for example, for all partial character string continuations, the partial character string continuation that can be preceded is added to the cumulative value Ctotal of the similarity of the partial character string itself. The degree of similarity C (n) is accumulated.

  Here, the case where the sum of the similarities of the partial character string continuation T is maximized in the case of the combination of the partial character string S0, the partial character string S1 + S2, and the partial character string S3 is shown.

As shown in FIG. 13, the partial character string sequence T0 to T3 includes the first partial character string S0 of the received mail 6, and therefore there is no previous partial character string sequence. In this case, the preceding partial character string continuation is not set, and the similarity C (Tn) × the number of partial character strings of each partial character string continuation Tn is set as the cumulative similarity Ctotal (Tn). That is, the similarity accumulation values Ctotal (T0), Ctotal (T1), Ctotal (T2), and Ctotal (T3) of T0, T1, T2, and T3 are calculated as follows.
Ctotal (T0) = C (T0)
Ctotal (T1) = C (T1) × 2
Ctotal (T2) = C (T2) × 3
Ctotal (T3) = C (T3) × 4

  Next, as shown in FIG. 14, since the partial character string sequence T4 to T6 includes the partial character string S1 at the beginning, only the partial character string sequence T0 including the partial character string S0 at the end is preceded. Can be made. For example, when the partial character string at the end of one partial character string continuation is Si, the first partial character string at the other partial character string continuation is S (i + 1). You just have to judge it. This i is a counter indicating the division order of the divided partial character strings. In this example, i is divided into four parts where i = 0 to 3.

When the preceding partial character string continuation is determined, the similarity cumulative value Ctoal is calculated for the partial character string continuation to be searched. In this case, the calculation of the similarity total value Ctoal is the similarity total value of successive partial character strings continuous + similarity of consecutive partial character strings × number of partial character strings. The preceding partial character string continuation of the partial character string continuations T4, T5, and T6 is T0, and the similarity accumulation values Ctotal (T4), Ctotal (T5), and Ctotal (T6) are as follows.
Ctotal (T4) = Ctotal (T0) + C (T4)
Ctotal (T5) = Ctotal (T0) + C (T5) × 2
Ctotal (T6) = Ctotal (T0) + C (T6) × 3

Similarly, in FIG. 15, the preceding partial character string continuation is specified for the partial character string continuations T7 and T8. As described above, the preceding partial character strings for the partial character string continuations T7 and T8 may be determined based on whether or not the included partial character string S is continuous with respect to the received mail 6. T4 is a target that can be preceded. Then, for these partial character string continuations T1 and T4, the similarity accumulation values Ctotal (T1) and Ctotal (T4) are compared, and the larger value is set as the partial character string continuation T7, T8 preceding partial character string continuation. . In this example, for example, Ctotal (T1)> Ctotal (T4), and the partial character string sequence T1 becomes the preceding partial character string sequence. Here, the similarity accumulation values Ctotal (T7) and Ctotal (T8) are calculated as follows.
Ctotal (T7) = Ctotal (T1) + C (T7)
Ctotal (T8) = Ctotal (T1) + C (T8) × 2

Similarly, as shown in FIG. 16, the partial character string continuation preceding the partial character string continuation T9 is, for example, partial character string continuation T2, T5, T7, and their similarity cumulative value Ctotal (T2). , Ctotal (T5), Ctotal (T7), the largest one is the partial character string continuation preceding the partial character string continuation T9. In this example, for example, partial character string continuation T5 is set. In this case, the cumulative value of the similarity of the partial character string sequence T9 is as follows.
Ctotal (T9) = Ctotal (T5) + C (T9)

  It should be noted that the accumulated similarity value information calculated for each partial character string sequence Tn is stored, for example, in the extraction information buffer 64 of the storage unit 58 and used for the search range setting process described later.

  Next, the most similar search range is determined for the entire received mail 6. In this case, for example, a combination of consecutive partial character strings that maximizes the cumulative value (total value) Ctotal of similarities is selected.

  In this partial character string continuous combination selection, for example, the partial character string continuous Tn including the partial character string S3 that constitutes the end of the sentence of the received mail 6 is selected. Therefore, the partial character string sequence T3, T6, T8, and T9 including the end of the mail is selected so as to maximize the similarity cumulative value Ctotal (Tn). In this example, for example, a case where Ctotal (T9) is the maximum is shown. In this case, as shown in FIG. 17A, a combination of consecutive partial character strings having the highest degree of similarity can be set by following the partial character string continuation preceding partial character string continuation T9.

  As a result, as shown in FIG. 17B, it is possible to set a search range based on combinations of partial character strings S0, S1 + S2, and S3, which are partial character string sequences T0, T5, and T9.

  (d) Reply sentence candidate extraction and display processing

  Next, the reply sentence candidate extraction process and display process based on the set search range will be described with reference to FIG. 18, FIG. 19, FIG. 20, FIG. 18 is a diagram illustrating an example of a reply message candidate extraction process, FIG. 19 is a diagram illustrating an example of a state where a reply message is selected, and FIG. 20 is a diagram illustrating an example of a display state of received mail. FIG. 22 is a diagram illustrating an example of a display state of divided received mails and extracted reply sentence candidates, and FIG. 22 is a diagram illustrating a display example of a reply mail creation screen. The configurations shown in FIGS. 18 to 22 are examples, and the present invention is not limited to these.

  A reply sentence candidate is extracted from the database 24 based on the partial character string continuation that is the set search range. In this reply sentence candidate extraction, for example, as shown in FIG. 18, a reply sentence candidate is presented for each partial character string sequence T that is set as a search range and has the maximum similarity. Here, the counter R is a counter of the set partial character string continuous T, and the counter i is a counter of the partial character string S combined with the partial character string continuous T. For example, in the search range setting shown in FIG. 17, the counter i of the partial character string S is set to i = 0-3. Further, the counter R may be set to R = 0 to 2 from the head of the sentence for the set combination of partial character strings T.

  The list of extracted reply sentence candidates may be temporarily stored in the extraction information buffer 64. The extracted reply sentence candidates are stored, for example, separated by a “|” symbol. Then, the extracted result is displayed on the display unit 50 of the mobile phone 40 using an appropriate interface.

  The user of the mobile phone 40 performs a selection operation on the presented reply sentence candidates and, for example, selects one sentence at a time as shown in FIG. Then, by combining the selected reply text information into one, a reply text 90 is created as shown in FIG. 19B.

  The mail screen display by the display unit 50 will be described. When the received mail 6 is displayed, a mail screen 92 is displayed on the display unit 50 as shown in FIG. The mail screen 92 is provided with, for example, a three-stage display area as a GUI (Graphical User Interface), and the upper stage includes a reception date, current time information, sender information, mail title information, and the like. An information display area 94 is configured. In the middle, a mail text display area 96 of the received mail 6 is configured. On the lower side, a function display area 98 in which a “menu” key, a “reply” key, a “select” key, and the like are allocated to the operation input unit 48 is configured. When an input operation is performed on the “reply” key 100 in the function display area 98, a reply mail creation function is activated.

  When the reply mail creation function is activated, the screen is switched to the reply mail creation screen 102 as shown in FIG. For example, the reply mail creation screen 102 is divided and displayed in a body display area 104, a received mail area 106, and a selection candidate area 108 from the upper side. The text display area 104 is an example of an area for displaying a reply sentence selected by the user or directly input, and displays a confirmed text. The received mail area 106 is an area for referring to a part of the received mail, and is displayed, for example, for each successive divided character string of the received mail 6 divided according to the principle described above. In the received mail area 106, for example, as shown in FIG. 8B, “Otsukase Sa-n” of the partial character string S0 is displayed. The selection candidate area 108 is an area for displaying a list of reply sentence candidates, and displays reply sentence candidates extracted based on the partial character strings displayed in the received mail area 106. In this selection candidate area 108, a selection cursor display 110 for the user to select with respect to the displayed candidates is displayed. For example, in response to the operation of the up / down operation buttons of the operation input unit 48, reply sentence candidates The selection target can be switched. Then, the selection cursor display 110 is aligned with the reply sentence candidate to be selected, and, for example, the reply sentence candidate is selected by operating a decision button or the like. As shown in FIG. The reply message is displayed on 104.

  When the reply message for the divided mail displayed in the received mail area 106 is confirmed, the display is switched to the display of the next divided mail. In conjunction with the display switching of the received mail area 106, the selection candidate area 108 is switched to a reply sentence candidate for the divided mail. As shown in FIGS. 21C and 21D, the reply message selection operation and the switching of the received mail are repeated until the reply text is selected for all the divided mails.

  In selecting a reply sentence candidate, for example, when there is no appropriate reply sentence or when it is not desired to reply, the “next sentence” option 112 displayed in the selection candidate area 108 is selected. Now you can compose an email that skips that sentence. When this option 112 is selected, for example, the next received mail 6 is displayed. In addition, when the user selects the “input end” option 114, for example, the user may return to the input screen and freely edit the text.

  When the reply message selection processing is completed for all received mails, for example, as shown in FIG. 22A, the mail screen is switched to the text display area 104 and the function display area 98. Will be displayed. When the “confirm” key 116 in the function display area 98 is operated, the mail transmission screen 117 is displayed. The mail transmission screen 117 includes, for example, an information display area 118 for displaying transmission destination information and title information, a text display area 104 for displaying a confirmed reply text, and a function display area 98. By operating the “Send” key 120 in the function display area 98, a reply mail is sent.

  Next, with reference to FIG. 23, FIG. 24, FIG. 25, and FIG. FIG. 23 is a flowchart illustrating an example of a reply mail creation process, FIG. 24 is a flowchart illustrating an example of a database search process, and FIGS. 25 and 26 are flowcharts illustrating an example of a partial character string continuous combination determination process. The processing contents, processing procedures, and the like shown in FIGS. 23 to 26 are examples, and are not limited to these. In FIGS. 25 and 26, A indicates a connector between flowcharts.

  This reply e-mail creation process is an example of the e-mail reply sentence extraction method of the present disclosure and its program, and in the search process for the database, the search range is set so as to maximize the similarity. Reply sentence candidates are extracted based on the search range.

  Therefore, the received mail 6 is divided into partial character strings S having a predetermined length based on, for example, punctuation marks, emoticons, pictograms, and the like (step S21). With respect to consecutive partial character strings composed of one or more partial character strings, the received sentence having the maximum similarity is searched from the received sentence / reply sentence database 24 (step S22), and the partial character string series having the maximum similarity is maximized. Is obtained (step S23). As shown in FIG. 17, this partial character string continuation combination is composed of one or a plurality of continuous partial character strings for the divided partial character string S, and this partial character string continuation retrieves an appropriate reply sentence. Search range.

  The number of consecutive partial character strings T that maximizes the determined similarity sum is set to M (step S24), and a counter R for the partial character string continuous T is set to R = 0 to initialize the reply sentence buffer. (Step S25). Then, a reply sentence candidate is extracted from the database 24 every partial character string sequence T (R).

  Therefore, if the counter R is smaller than the number M of partial character string continuations T (YES in step S26), the partial character string continuation T ( For R), a reply sentence candidate is extracted from the database 24 and displayed on the reply mail creation screen 102 (step S27). The reply sentence selected by the user is added to the reply sentence buffer (step S28). This reply sentence buffer holds, for example, reply mail information being created. For example, the extracted information buffer 64 may be used. Then, this process is repeated while R <M while incrementing the counter R (step S29). When the reply sentence is confirmed for all the partial character string sequences T (NO in step S26), the reply sentence character string stored in the reply sentence buffer is output to complete the reply mail (step S30).

  Next, the database search process for partial character string continuation in step S22 executes, for example, the process shown in FIG.

  In this database search process, for example, as shown in FIG. 12, a received sentence with the maximum similarity is received / replied with respect to a large number of consecutive partial character strings S composed of one or a plurality of consecutive partial character strings S. Search from the sentence database 24. Therefore, as shown in FIG. 24, if the number of the divided partial character strings S is N (step S51), for example, if the counter of the partial character string S set based on the division order for the received mail is i, i = Set to 0 to initialize the counter (step S52).

  Next, in order to search the database for all of a large number of partial character string sequences T set by a combination of one or a plurality of consecutive partial character strings S, the counter i reaches the number N of divided partial character strings S. , I is incremented (YES in step S53). In this case, a counter j for setting combinations for all the partial character strings S is set for the counter i, and the counter j is set to i + 1 (step S54). That is, the counter i indicating the division order of the partial character string S is incremented by 1 from i = 0, and the counter j is incremented by 1 from j = 1.

  The database search process and the similarity calculation process are repeated in order from the counter j = 1 while the counter j <(N + 1) (YES in step S55). Therefore, the database is searched for the partial character string continuation from the partial character string Si to the partial character string Sj-1, and the data having the maximum similarity and the similarity C are obtained (step S56). Specifically, first, when the counter i = 0, referring to FIG. 12, it indicates that the detection process of the partial character string continuation T including the partial character string S0 is performed. That is, when i = 0 and j = 1, a partial character string continuous T0 composed only of the partial character string S0 is indicated. Further, when the counter j is incremented (step S57), if j = 3, the partial character string sequence T3 constituted by S0 + S1 + S2 is obtained. Then, the similarity C is calculated for each partial character string sequence. Then, the counter i is incremented (step S58). When i = 1, j is counted from j = 2, and by incrementing j, the partial character string sequence T of S1, S1 + S2,. Is done.

  When the database search process and the similarity calculation process for all the combinations are performed on the divided partial character string S (NO in step S53), this process ends.

  Next, the partial character string continuous combination determining process that maximizes the similarity in step S23 (FIG. 23) may be performed, for example, as shown in FIG.

  The partial character string continuous combination determination processing is performed based on, for example, the Viterbi algorithm. This process includes a process (process F1) for obtaining a preceding partial character string continuation that maximizes the total similarity for each partial character string continuation. In addition, the process of obtaining the optimum combination of partial character strings (the maximum similarity is maximized) is obtained by tracing the preceding partial character string continuation so that the similarity total increases from the last partial character string continuation (Process F2). )including.

  Therefore, in process F1, the number of partial character string continuations T is set to X (step S61). The number X may be calculated from a combination of partial character strings S obtained by dividing the received mail as shown in FIG.

  A counter indicating the combination order of the partial character string sequences T is set as I, and the counter I is initialized as I = 0 (step S62). This counter I may be set, for example, in the same way as the counter n of the partial character string continuation T shown in FIG. 12, and from the partial character string Si set in step S56 of FIG. 24 described above to the partial character string Sj−1. You may set according to the extraction order. The counter I is repeatedly incremented up to the number X of partial character string continuations T (YES in step S63).

  Here, the similarity cumulative value Ctotal (T (I)) of the partial character string sequence T (I) is set to 0 (step S64). Also, a counter J is set for detecting a partial character string continuation that is a preceding partial character string continuation with respect to a partial character string continuation T (I) of the counter I to be inspected. Initialization is performed (step S65). The counter J counts a partial character string sequence T having the same contents as the counter I. The counter J is also incremented repeatedly up to the number X of partial character string continuations T (YES in step S66).

  The partial character string sequence T (J) determines whether it can precede the partial character string sequence T (I) (step S67), and if it can precede (YES in step S67), the similarity is determined. It is determined whether or not the cumulative value Ctotal (T (J)) is larger than Ctotal (T (I)) (step S68). If Ctotal (T (J))> Ctotal (T (I)) (YES in step S68), Ctotal (T (J)) is substituted for Ctotal (T (I)) and substring continuous T ( The preceding partial character string sequence of I) is set to T (J) (step S69).

  Here, whether or not it precedes, for example, the partial character string continuous T (J) includes a partial character string preceding the partial character string S included in the partial character string continuous T (I), and the partial character string. It may be determined that the continuous T (J) and the partial character string continuous T (I) are continuous. For example, a description will be given based on FIG. In the counters I and J set in the same manner as the counter n, first, when I = 0, that is, when T (0) is a search target, even if the counter J is incremented, the portion preceding T (I) There is no character string sequence T (J) (NO in step S67). Similarly, when I = 1 to 3, there is no preceding partial character string T (J).

  For example, when I = 4, the partial character string T (J) that can precede the target partial character string T (4) indicates T (0) (YES in step S67). . In this case, since Ctotal (T (I)) = 0 (step S64), it is determined that Ctotal (T (J)) is larger than Ctotal (T (I)) (YES in step S68). ). As described above, the processing of step S68 and step S69 is processing for selecting the one having the largest accumulated similarity value from the preceding partial character string continuation T (J), and incrementing the counter J. By doing so, it is possible to extract the preceding partial character string continuity having a large similarity cumulative value for each set I.

  When the preceding partial character string sequence T (J) with the maximum similarity is determined, the similarity C () of the partial character string sequence T (I) of the counter I is added to the similarity cumulative value Ctotal (T (I)). T (I)) is added to obtain a cumulative similarity Ctotal (T (I)) of the partial character string sequence T (I) (step S71). Here, “Ctotal (T (I)) + = C (T (I))” shown in step S71 is “Ctotal (T (I)) = Ctotal (T (I)) + C (T (I)))”. ". Then, the counter I is incremented to calculate the similarity cumulative value Ctotal (T (I)) for all I (step S72).

  Next, when the counter I ≧ X (NO in step S63), the process proceeds to the process of determining the optimum combination of partial character strings as process F2. That is, in the process F1, the calculation result of the partial character string sequence T (I) having the maximum similarity accumulation value is referred to for all I (step S73). In this combination determination process, as described above, the partial character string sequence T having the maximum similarity is determined from the end of the received mail 6. Therefore, the search result of the partial character string sequence T (I) is output (step S74), and the preceding partial character string until the partial character string sequence T (I) includes the partial character string S0 at the beginning of the received mail 6. Following the continuation T, extraction is performed, and the extracted partial character string continuation T (I) is output as a search result (steps S75 and S76). This process is continued until there is no preceding partial character string continuation.

  With such processing, a combination of partial character strings that maximizes the similarity can be set as a search range, and the detection process of step S27 (FIG. 23) may be performed based on this search range.

  In this embodiment, the Viterbi algorithm is used to determine the combination of partial character strings. However, the combination of partial character strings that maximizes the similarity may be determined by another method. .

  According to such a configuration, it is possible to search the database in an optimum unit for the received mail and obtain the reply sentence candidate. And since a user selects a reply sentence from the extracted reply sentence candidates and connects them to create a reply mail, the burden on the user can be reduced. In addition, by setting the search range so that the mail body and the stored received sentence candidate are the closest contents, appropriate reply sentence candidates can be extracted, and a response mail is created according to the contents of the mail body can do.

  [Third Embodiment]

  Next, a third embodiment will be described with reference to FIGS. 27 and 28. FIG. FIG. 27 is a diagram illustrating a configuration example of a reply mail creation function according to the third embodiment, and FIG. 28 is a diagram illustrating a configuration example of a received text / reply text database. The configuration shown in FIGS. 27 and 28 is an example, and the present invention is not limited to this. 27 and 28, the same parts as those in FIG. 5 and the like are denoted by the same reference numerals, and the description thereof is omitted.

  The information processing apparatus 2 is an example of the information processing apparatus of the present disclosure, and sets an appropriate search range for the received mail 6 and automatically extracts a reply sentence candidate based on the search range. A mail creating apparatus 131 having an automatic reply function for determining a reply sentence is configured. This mail creation device 131 corresponds to the mail creation device 4 described above. The automatic reply function of the mail creation device 131 is set for the contents included in the received mail 6 in the set search range, for example, a reply to a greeting or the like that does not necessarily require the user's judgment. Automatically determines the response text. The information processing apparatus 2 includes, for example, a received text / reply text database 130 that stores automatic reply availability information 132, an automatic selection / user selection switching section 134, a reply text automatic selection section 136, and the like. In addition, as described above, the mail dividing unit 20, the received text selecting unit 10, the search unit 22 including the search range setting unit 14, the reply text extracting unit 16, the reply text combining unit 26, the reply text candidate display unit 28, the selection What is necessary is just to provide the input part 30 grade | etc.,.

  The received sentence / reply sentence database 130 stores a received sentence candidate to be searched and a reply sentence candidate associated with the received sentence candidate for one or a plurality of continuous partial character strings set as a search range. This is an example of storage means, and corresponds to the received sentence / reply sentence database 24 described above. In this received sentence / reply sentence database 130, for example, as shown in FIG. 28, a reply sentence candidate field 82 that stores reply sentence candidates in association with the received sentence field 80 is set, and automatic reply availability information An automatic reply availability column 138 including 132 is set.

  The automatic reply availability information 132 is information for setting whether or not an automatic reply is possible according to the content of the received text, for example. The setting contents of the automatic reply availability column 138 may be set by the user, for example. Received sentences that enable automatic replies, such as greetings such as “Otsukare!”, Which are likely to be placed at the beginning of the sentence, and are unlikely to ask for a reply such as “No reply needed” The thing should just be set. As described above, when a received sentence that is highly likely not to require the user's judgment is detected in creating a reply sentence, an automatic reply using a reply sentence candidate is enabled. In addition, for example, when a received sentence that requires judgment by the user is searched, such as “Let's go to see a movie!” Or “Tomorrow's free time?”, Automatic reply is set to be impossible.

  In addition, for example, it may be possible to switch whether or not automatic reply is possible in accordance with the sender of the received mail.

  The automatic selection / user selection switching unit 134 refers to the set automatic reply availability information 132 and automatically selects a reply sentence from reply sentence candidates in the database 130 or presents a reply sentence candidate to the user for selection. It is an example of a means for switching whether to prompt. As described above, for the received mail 6, the search unit 22 sets a search range in which partial character string sequences T having the highest similarity are combined. Then, reference is made to whether or not the automatic reply availability information 132 is set in the received sentence or reply sentence candidate searched based on this search range. When automatic reply impossible is set, the reply sentence candidate in the received sentence / reply sentence database 130 is displayed on the reply sentence candidate display unit 28 as described above. If automatic reply is possible, an instruction to switch to automatic selection of reply text is output to the reply text automatic selection unit 136. This automatic selection / user selection switching function is realized by, for example, a mail creation program stored in the program storage unit 60.

  The reply sentence automatic selection unit 136 is an example of means for automatically selecting one or a plurality of reply sentences from the reply sentence candidate column 82 of the received sentence / reply part database 130. When receiving an instruction to switch to automatic selection from the automatic selection / user selection switching unit 134, for example, a random number is used to select a reply sentence from the reply sentence candidate column 82 associated with the received received sentence. The selected reply message is notified to the reply message combining unit 26 side, and a reply mail 31 is created.

  Note that the reply text automatic selection unit 136 is not limited to the case of using a random number. For example, the reply text candidate selection order may be switched according to the storage order of reply text candidates, and the user associates with a mail recipient or the like in advance. Reply sentence candidates may be set.

  Then, for each search range, the reply sentences selected by the automatic selection or the user selection are combined by the reply sentence combining unit 26, and a reply mail 31 is created.

  Next, FIG. 29 is referred to for the reply mail creation process. FIG. 29 is a flowchart illustrating an example of a reply mail creation process. The processing content or processing procedure shown in FIG. 29 is an example, and the present invention is not limited to this.

  This reply mail creation process is an example of the e-mail reply sentence extraction method of the present disclosure and its program. As described above, an optimal search range is set for the received mail, and the corresponding reply Extract sentence candidates. The reply mail creation process includes a reply text automatic selection process for each set search range.

  Therefore, as a search range setting for the received mail, the received mail is divided (step S91), the partial character string continuous database search (step S92), the partial character string continuous combination determination that maximizes the similarity is determined (step S93), The number of consecutive partial character strings may be set to M (step S94) in the same manner as steps S21 to S24 shown in FIG.

  Next, the process proceeds to a reply sentence candidate selection process for each search range based on the set of consecutive partial character strings. In this case, the counter R of the partial character string continuation T set in the search range is initialized (step S95), and this R is incremented (step S102), thereby selecting a reply sentence for each search range. Therefore, it is determined whether or not the retrieved reply sentence candidate is a reply that requires user judgment (step S97). When the user's judgment is not necessary (NO in step S97), that is, when automatic reply is possible, one sentence is automatically selected from the reply sentence candidates (step S98), and the selected reply sentence is selected. If it is added to the reply sentence buffer (step S99), or if user judgment is necessary (YES in step S97), the reply sentence candidate is presented to the user as described above (step S100), and the user selects it. The reply sentence that has been added is added to the reply sentence buffer (step S101).

  Steps S95, S96, and S100 to S103 may be performed in the same manner as steps S25 to S30 in FIG. 23, for example.

  According to such a configuration, it is possible to reduce the number of search targets for the user to select a reply sentence, to reduce the labor for creating a reply mail, and to improve convenience. In addition, even in the configuration in which the reply text is automatically selected in this way, the search range is set so that the mail text and the stored received text candidate are the closest contents, so that an appropriate reply text candidate can be extracted. Can create an appropriate response email.

[Fourth Embodiment]

  Next, a fourth embodiment will be described with reference to FIGS. 30 and 31. FIG. FIG. 30 is a diagram illustrating a configuration example of a reply mail creation function according to the fourth embodiment, and FIG. 31 is a diagram illustrating a display example of the created reply mail. The configuration shown in FIGS. 30 and 31 is an example, and the present invention is not limited to this.

  This information processing apparatus 2 is an example of the information processing apparatus of the present disclosure. As described above, an appropriate search range is set for the received mail, and reply sentence candidates are extracted based on the search range. A mail creation device 141 is provided. The information processing apparatus 2 is configured to display a reply sentence candidate extracted according to the search range and a received sentence corresponding to the search range in the reply mail to be created. In other words, the information processing apparatus 2 has a function of clearly indicating to which part of the received mail the prepared reply sentence is.

  For example, the mail creation device 141 may include a display control unit 140, a reply sentence generation unit 142, and the like as shown in FIG. In addition, as described above, the mail dividing unit 20, the received text selecting unit 10, the search unit 22 including the search range setting unit 14, the reply text extracting unit 16, and the like may be provided. In addition, a reply sentence candidate display unit 28, a selection input unit 30, and the like are connected to the mail creation device 141.

  The display control unit 140 associates the reply sentence candidate searched based on the set search range with the partial character string continuation T of the received mail 6 to be searched and displays it on the reply sentence candidate display unit 28 side. It is an example. The search unit 22 determines a combination of partial character string sequences T having the highest similarity for the received mail 6 divided into predetermined partial character strings S, and sets this as a search range. Then, the display control unit 140 receives the reply sentence candidate extracted from the received sentence / reply sentence database 24 based on the search range, the received sentence, and the like from the reply sentence extraction unit 16, and displays the reply sentence candidate display unit 28 side. Send to.

  Further, the reply sentence generation unit 142 attaches an identification symbol for identifying the received sentence to the divided mail to be searched, and associates the divided mail received from the search part 22 with the reply sentence information. It is an example of a means for creating a reply mail.

  This identification symbol is an example of a means for visually recognizing that it is a part of the received mail 6 corresponding to the extracted reply sentence. As shown in FIG. 31, on the reply mail screen 144, for the divided mail, For example, an identification symbol 146 of “>” is attached to the beginning of the sentence. In the next paragraph of the divided mail, a reply sentence column 148 searched and selected for the divided mail is set. This reply message column 148 has no identification symbol. Then, for example, the divided mails with the identification symbol 146 and the reply sentences corresponding to the divided mails are displayed one by one to create a reply mail.

  The identification information is not limited to being added to the divided mail, and for example, an identification symbol may be added to the reply sentence candidate side. In this embodiment, the display control unit 140 also presents the divided received mail 6 that has been searched for reply candidates to the user side. However, the present invention is not limited to this. May not display the divided received mail 6. When the reply message is generated, the reply mail with the divided mail with the identification symbol and the selected reply text may be displayed.

  Next, with reference to FIG. FIG. 32 is a flowchart illustrating an example of a reply mail creation process. The processing content or processing procedure shown in FIG. 32 is an example, and the present invention is not limited to this.

  This reply mail creation process is an example of the e-mail reply sentence extraction method of the present disclosure and its program. As described above, an optimal search range is set for the received mail, and the corresponding reply Extract sentence candidates. In addition, for each set search range, a divided mail with an identification symbol and the reply message are displayed together.

  Therefore, as a search range setting for the received mail, the received mail is divided (step S111), the partial character string continuous database search (step S112), the partial character string continuous combination determination that maximizes the similarity (step S113), The number of consecutive partial character strings is set to M (step S114), and the initialization of the counter R (step S115) may be performed in the same manner as steps S21 to S25 shown in FIG.

  In selecting a reply sentence, the partial character string sequence T (R) is quoted in the received mail 6 and added to the reply sentence buffer (step S118). In addition, steps S116 to S117 and steps S119 to S121 may be performed in the same manner as steps S26 to S30 in FIG.

  According to such a configuration, it is possible to generate a reply mail that is easier to understand for the reply partner by causing the contents of the received mail text and the reply text selected for the mail to be written together. In addition, by setting the search range so that the mail body and the stored received sentence candidate are the closest contents, appropriate reply sentence candidates can be extracted, and a response mail is created according to the contents of the mail body can do. By extracting appropriate reply sentence candidates according to the contents of the mail body, the user can create a response mail with only a selection operation, reduce the labor of character input operation, and create a quick response mail. It can be carried out. Furthermore, since reply sentence candidates for the contents are extracted and displayed along the text described in the mail text, it is possible to respond to the contents described by the sender of the mail without omission.

[Fifth Embodiment]

  Next, a fifth embodiment will be described with reference to FIG. 33, FIG. 34, and FIG. FIG. 33 is a diagram showing a configuration example of a reply mail creation function according to the fifth embodiment, FIG. 34 is a diagram showing a configuration example of a received text / reply text database, and FIG. 35 is a process for extracting and replacing specific information It is a figure which shows an example. The configurations shown in FIGS. 33 to 35 are examples, and the present invention is not limited to these.

  This information processing apparatus 2 is an example of the information processing apparatus of the present disclosure, and extracts an appropriate search range setting function for received mail and specific information included in the received mail, A mail creation device 151 including a function of creating a reply mail that incorporates the extracted specific information is provided. This specific information is, for example, date / time information, place name, place name information, etc. included in the received mail 6. These specific information, for example, date, place, etc. vary in content, but the description target is the same, and information that can specify the position to be described in the received mail document and the reply mail document It is.

  Therefore, a rule is set for detecting specific information such as date and place included in the received mail 6, and this specific information is extracted. In the received text / reply text database 24, reply text candidates that specify the position where the specific information is inserted are stored. In creating a reply mail, the extracted specific information is replaced with a specific position for the selected reply sentence. For example, a replacement tag that can identify the type of specific information that can be inserted may be set at the replacement position of the specific information.

  The mail creation device 151 includes, for example, a specific information extraction unit 150, a specific information storage unit 152, a specific information replacement processing unit 154, a reply sentence generation unit 156, and the like. Further, as described above, in order to set an appropriate search range, the mail dividing unit 20, the received sentence / reply sentence database 24, the reply sentence extracting unit 16, and the like are provided. A reply sentence candidate display unit 28 and a selection input unit 30 are connected to the mail creation device 151.

  The specific information extracting unit 150 is an example of means for extracting specific information for each partial character string obtained by dividing the received mail 6. For example, according to a preset extraction rule, the “specific date”, “location”, “ "Location name" information and the like are extracted. Then, the extracted specific information is transmitted to the specific information storage unit 152. As this extraction rule, for example, as a date and time detection rule, a character string representing a date and time such as “Today”, “Tomorrow”, “Tomorrow”, “Tonight”, “Next week”, “Next month”, “Monday” to “Friday” is prepared as a table. Can be verified. Further, it can be realized by detecting a character string in which “month” follows a number from 1 to 12 or a character string in which “day” follows a number from 1 to 31. The location detection rule may be prepared by collating the main station names and landmarks as a table, for example. This extraction process may use, for example, a morpheme analysis method that analyzes a part of speech of a target word using an extraction rule.

  The specific information storage unit 152 is an example of a unit that stores the extracted specific information, and may be configured in the data storage unit 62 (FIG. 7), for example. In addition, the specific information storage unit 152 provides specific information corresponding to the tag included in the reply text when creating the reply text.

  The specific information replacement processing unit 154 replaces the specific information included in the partial character string with a tag corresponding to the type, and if a tag is included in the selected reply sentence, the tag Is an example of a means for replacing with the specific information stored in the specific information storage unit 152. As described above, when a reply sentence candidate including a tag is extracted from the database 24 based on the set partial character string continuation as an appropriate search range in the search unit 22, the corresponding specific information Is replaced. This specific information replacement processing function is realized by, for example, a mail creation program stored in the program storage unit 60 (FIG. 7).

  The reply sentence generation unit 156 is an example of a means for creating a reply mail by using a reply sentence in which the specific information replacement processing unit 154 replaces a tag with a date and time or a place name.

  As described above, the search unit 22 searches the partial character string after the replacement of the specific information and the received sentence candidate in the database, and sets the combination of partial character strings that have the highest similarity. Then set the search range.

  In the received text / reply text database 24, for example, as shown in FIG. 34, a tag 158 is set at a position where specific information such as the date and time in the mail is inserted in advance in the received text and reply text candidates. Therefore, in the received sentence similarity calculation process, the similarity between the partial character string in which the specific information is replaced with the tag and the received sentence in which the tag is set is calculated. Also, in the extraction of reply sentence candidates, reply sentences with tags set are extracted.

  The specific information extraction process, tag replacement process, database search process, and reply sentence candidate extraction process for the partial character string are executed, for example, as shown in FIG.

  Here, for example, a case where only one partial character string is generated when the received mail 6 is divided, and this partial character string S is used as a search range is illustrated.

  As shown in FIG. 8A and FIG. 8B, as a result of dividing the received mail 6, for example, one partial character string S0 “Do not drink around Shibuya on Saturday?” Is obtained (A in FIG. 35). ). For this partial character string S0, in accordance with the extraction rules for specific information, for example, as shown in FIG. 35B, a display 160 of “Saturday” is displayed as date specific information and a display 162 of “Shibuya” is specified as location specific information. Extracted and replaced with a date tag 164 and a place tag 166, respectively. As a result, as shown in C of FIG. 35, the partial character string S0 becomes “Do not drink around <location> at <date and time>?”. Then, when this is searched in the database 24, a received sentence “Let's drink at <location> at <date and time>” is searched as a received sentence with a high degree of similarity.

  For example, as shown in FIG. 35D, the reply sentence candidate related to the received received sentence includes a tag 158 or the like, such as “I'm sorry, <date and time> is inconvenient.” There is something. If the extracted reply sentence candidate includes a tag 158, as shown in E of FIG. 35, the display 160 of “Saturday” is replaced as the extracted specific information, for example, “I ’m sorry, "Saturday is inconvenient."

  Then, as shown in F of FIG. 35, the user selects a reply sentence in which the specific information is replaced, and creates a reply mail as shown in G of FIG.

  Next, with reference to FIGS. 36 and 37, the reply mail creation process for the received mail including the specific information will be described. 36 and 37 are flowcharts showing an example of reply mail creation processing. The processing contents and processing procedures shown in FIGS. 36 and 37 are examples, and the present invention is not limited to these. In FIGS. 36 and 37, B indicates a connector between flowcharts.

  This reply mail creation process is an example of an e-mail reply sentence extraction method of the present disclosure and a program thereof, which extracts specific information included in the received mail and uses the specific information part as a tag with common contents. Substitution is performed, and search range setting and reply sentence candidate extraction processing are performed.

  The received mail 6 is divided into partial character strings for each predetermined unit (step S131), and a character string corresponding to specific information such as date and time is detected in this partial character string. Then, the character string is replaced with a tag indicating date and time information (step S132). Further, a character string corresponding to specific information such as a location is detected in the partial character string, and this character string is replaced with a tag indicating the location information (step S133).

  As described above, with regard to the partial character string replaced with various tags for the specific information, as described above, the database search process (step S134) and the setting process of the combination of consecutive partial character strings that maximize the similarity degree (step S135) ) To set the search range. The processes in steps S134 to S138 in FIG. 36 may be performed in the same manner as steps S22 to S26 in FIG. 23, for example.

  When there is a tag indicating the date and time in the reply sentence candidate extracted for each partial character string continuation T (R) according to the set counter R for counting the partial character string continuations (YES in step S139), for example, a specific information storage unit The date and time specific information 152 is replaced with this tag (step S140). If there is a tag indicating the location in the extracted reply sentence candidates (YES in step S141), the location specific information in the specific information storage unit 152 is replaced with this tag (step S142). Then, the reply sentence candidate subjected to the replacement process is presented to the user (step S143), and a reply sentence selection process is performed. The processing in steps S143 to S146 may be performed corresponding to steps S27 to S30 in FIG.

  In this embodiment, the date information and the location information are exemplified as the specific information, but the specific information is not limited thereto. For example, other specific information such as a person's name or a specific word set by the user may be set in the specific information, and the replacement process using a tag may be performed. Further, the process of returning the specific information to the replacement tag is not limited to the case where the specific information extracted from the received mail is used. For example, replacement data may be set in the specific information storage unit 152 in advance. In addition, the user may directly input the tag portion.

  According to such a configuration, it is possible to search from the database for mails with similar contents that differ only in date and time, and it is possible to extract appropriate reply sentence candidates for more various mails. . In the search range setting process, for example, it is possible to prevent a similarity calculation value from being lowered for a received sentence candidate in which specific information such as date and place is different but the contents of the received sentence are the same. Can do. In addition, even if there are many matching characters, words, codes, etc. included in the received text candidate for the received mail, it is possible to prevent a received text candidate that does not match the contents of the document from being selected. The detection range setting accuracy can be increased.

[Sixth Embodiment]

  Next, a sixth embodiment will be described with reference to FIGS. 38, 39, and 40. FIG. FIG. 38 is a diagram illustrating a configuration example of a reply mail creation function according to the sixth embodiment, FIG. 39 is a diagram illustrating a configuration example of an expression replacement character string table, and FIG. 40 is a diagram illustrating expression replacement processing and database search processing. It is a figure which shows an example. The configurations shown in FIGS. 38 to 40 are examples, and the present invention is not limited to these.

  This information processing apparatus 2 is an example of the information processing apparatus of the present disclosure, and when the words included in the received mail 6 are synonymous with the received sentence candidates stored in the received sentence / reply sentence database 24, An e-mail creation device 171 that performs search processing by replacing the words is provided. The various expressions contained in the received mail 6 are replaced with, for example, expressions and terms that can be easily searched by the database 24, thereby improving the search accuracy and setting an appropriate search range. According to the contents of the received mail 6 Enables extraction of reply sentence candidates. This expression or term replacement may be any expression that normalizes or unifies expressions for abbreviations, terms used in a specific field, etc., for words that indicate the same content, for example.

  This mail creation device 171 may include, for example, an expression replacement unit 170, an expression replacement character string table 172, and the like. Further, as described above, the mail dividing unit 20, the search unit 22, the reply message extracting unit 16, the reply message combining unit 26, and the like may be used.

  The expression replacement unit 170 extracts words, words, etc. that are synonymous with words and expressions stored in the database as received sentence candidates and are different from the partial character strings of the received mail 6, and store them in the database. It is an example of a means for replacing with stored words and expressions. In the extraction of words included in the partial character string, for example, the expression replacement character string table 172 is used, and when it is the same as the word included therein, it is extracted as a word to be replaced. Then, it is replaced with a word or expression corresponding to the expression replacement character string table 172 and inserted into the partial character string. The expression replacement unit 170 is executed by, for example, a mail creation program stored in the program storage unit 60 (FIG. 7).

  The expression replacement character string table 172 is a storage unit that stores words, expressions, and words that are synonymous with and displayed differently in terms of words, expressions, and words included in the received sentence candidates of the received sentence / reply sentence database 24. It is an example. The expression replacement character string table 172 includes, for example, a conversion destination field 175 associated with the conversion source field 174 as shown in FIG. As words included in the conversion destination column 175, words included in the received sentence candidate stored in the received sentence / reply sentence database 24 are set. The conversion source column 174 is a synonymous expression or word for the word set in the conversion destination column 175, for example, a word in an expression format often used in e-mail or the like. The expression replacement character string table 172 is stored in, for example, the data storage unit 62 (FIG. 7) of the storage unit 58.

  In addition, as described above, in the mail creation device 171, an appropriate search range is set by the search unit 22 for the partial character string whose expression is replaced, and the reply sentence extraction unit 16 uses the search range based on the search range. A reply sentence candidate is extracted and presented on the reply sentence candidate display unit 28. Then, the reply mail 31 is created by combining the selected reply sentence candidates.

  The expression replacement process for the partial character string and the database search process for the partial character string subjected to the replacement process are executed as shown in FIG. 40, for example.

  In this processing example, for example, the received mail 6 has a single divided partial character string. As a result of dividing the received mail 6, for example, “How about BBQ together?” Is obtained as the partial character string S0. When a search process using the expression replacement character string table 172 shown in FIG. 39 is performed on this partial character string, for example, as shown in A of FIG. 40, a display 176 of “BBQ” and “How? ”Is extracted.

  Then, as shown in FIG. 40B, the expression replacement unit 170 replaces the “BBQ” display 176 with the “barbecue” display 180 based on the expression replacement character string table 172, and displays the “how?” Display 178. Is replaced with a display 182 of “How is it?”.

  Then, as shown in FIG. 40C, a search process for the received sentence / reply sentence database 24 is executed based on the partial character string S0 subjected to the replacement process. The words in the received text column 80 of the received text / reply text database 24 correspond to the words set as the conversion destination in the expression replacement character string table 172. As a result of this search processing, reply sentence candidates are extracted as shown in FIG. 40D, and a reply mail is created based on the reply sentence selected by the user for the reply sentence candidates.

  In this example, with the expression “How about BBQ together?” Shown in the partial character string S0 before conversion, the equivalent content is set for “How about BBQ together” set in the received text / reply database 24. In spite of the expression, the similarity of the character string becomes low, and there is a possibility that it is not searched. Therefore, by performing the expression conversion process, the degree of similarity is increased with respect to the sentence in the database, and the possibility of being searched can be increased.

  Next, with reference to FIG. 41, a reply mail creation process including an expression replacement process will be described. FIG. 41 is a flowchart illustrating an example of a reply mail creation process. The processing content, processing procedure, and the like illustrated in FIG. 41 are examples and are not limited to these.

  This reply mail creation process is an example of an e-mail reply sentence extraction method and program thereof according to the present disclosure, and includes conversion processing for a character string that matches the expression replacement character string table 172 in the partial character string.

  Therefore, as described above, the received mail 6 is divided into partial character strings for each predetermined unit (step S151), and if this partial character string includes a character string that matches the expression replacement character string table 172, A replacement process is performed (step S152). Then, a database search or a partial character string continuation setting process is performed on the partial character string subjected to the replacement process, a search range is set, and a reply sentence candidate is extracted. The processing from step S153 to step S161 corresponds to step S22 to step S30 in FIG.

  In the expression replacement character string table 172, it is possible to set various character strings such as words and sentence ending expressions that have the same meaning but can be expressed in various ways. However, for example, when replacement is performed by simple character string matching, or when a character string that is too short or a similar character string is conceivable, it is necessary to avoid inappropriate replacement. Therefore, for example, after morphological analysis is performed and the words constituting the partial character string are specified, the replacement may be set in units of words.

  According to such a configuration, by unifying the various expressions included in the received mail, the possibility of searching mails having the same meaning but different expressions from the database is increased. On the other hand, an appropriate reply sentence candidate can be extracted.

[Seventh Embodiment]

  Next, the seventh embodiment will be described with reference to FIGS. 42 and 43. FIG. FIG. 42 is a diagram illustrating a configuration example of a reply mail creation function according to the seventh embodiment, and FIG. 43 is a diagram illustrating a configuration example of a received text / reply text database. The configuration shown in FIGS. 42 and 43 is an example, and the present invention is not limited to this.

  This information processing apparatus 2 is an example of the information processing apparatus of the present disclosure, and as described above, in addition to the search range setting function for received mail, the reply sentence candidate is selected according to the reply sentence selected immediately before. A mail creation device 191 including a function for changing the display order is provided. In this mail creation function, a reply sentence that is highly likely to be selected by the user is displayed higher on the list of reply sentence candidates extracted based on the set search range. Therefore, in the database storing the reply sentence candidates, the reply sentences are classified based on the attribute information for each reply sentence, and the reply sentence candidates are displayed based on the classification information that is highly likely to be selected by the user.

  The mail creation device 191 includes, for example, a received sentence / reply sentence database 190, a reply sentence candidate display adjustment unit 193, and the like. Further, as described above, the mail division unit 20, the search unit 22 for setting the search range, the reply sentence extraction unit 16, and the reply sentence combination unit 26 are provided. In addition, the information processing device 2 includes a reply sentence candidate display unit 28 and a selection input unit 30, and is connected to the mail creation device 191.

  The received sentence / reply sentence database 190 is an example of a means for storing received sentence candidates corresponding to partial character strings obtained by dividing the received mail 6 and reply sentence candidates associated with the received sentence candidates. Corresponding to the received / reply text database 24. The received sentence / reply sentence database 190 includes, for example, an attribute information storage unit 192 that stores attribute information for classifying reply sentence information.

  In the received text / reply text database 190, as shown in FIG. 43, for example, an attribute column 194 is set for each reply text candidate, and classification is performed based on the attribute information. This attribute column 194 is an example of the attribute information storage unit 192. As this attribute information, for example, an “accept” column 196, a “reject” column 198, a “hold” column 200, and an “no attribute” column 202 are set, and reply sentence candidates are classified for each column. In this attribute information, for example, “accept”, “reject”, and “hold” attributes are set in accordance with the content of a reply for a received sentence that is a request for an answer of acceptance or rejection. In addition, “no attribute” is set as an attribute of a reply sentence with respect to a received sentence that does not require a response.

  In addition, the kind of attribute shown here is an example, Comprising: It is not limited to this.

  As described above, the reply sentence candidate display adjustment unit 193 classifies the reply sentence candidates extracted based on the appropriate search range into the same attribute based on the reply sentence attribute information selected in the past. This is a means for preferentially displaying the reply sentence candidates to be displayed on the reply sentence candidate display unit 28. This reply message candidate display adjustment unit 193 is connected to the selection input unit 30, for example, and stores the attribute information of the selected reply sentence when the user selects a reply sentence from the reply sentence candidates. In the next response sentence candidate display, the display order is changed based on the attribute information.

  Next, with reference to FIG. 44, a reply mail creation process including display adjustment of a reply sentence candidate based on attribute information will be described. FIG. 44 is a flowchart illustrating an example of a reply mail creation process. The processing content, processing procedure, and the like shown in FIG. 44 are merely examples, and the present invention is not limited to these.

  This reply e-mail creation process is an example of an e-mail reply sentence extraction method and program thereof according to the present disclosure, and includes an optimum search range setting process and a reply sentence candidate display adjustment process based on attribute information.

  Therefore, the received mail 6 is divided into predetermined partial character strings (step S191), and a database search for this partial character string and a partial character string continuous combination setting process that maximizes the similarity are performed. Steps S191 to S195 in FIG. 44 correspond to the processes shown in steps S21 to S25 in FIG.

  The presentation order priority information of reply sentence candidates is initialized to “no attribute” (step S196). This presentation order priority information is the setting information of the presentation order of reply sentence candidates extracted from the database 190. In the case of “no attribute”, the presentation order priority information may be presented according to the order stored in the database 190. Based on the set partial character string continuation T (R), when the reply sentence candidate extraction process is repeated, the reply sentence candidate order is changed and presented to the user according to the set presentation order priority information. (Step S198).

  By adding the reply text selected by the user to the reply text buffer (step S199), when attribute information is set in the selected reply text, the attribute information is set as presentation order priority information (step S200). ). Then, in the reply sentence extraction process for the next partial character string series T (R), the presentation order of reply sentence candidates is changed based on the presentation order priority information set here.

  Reference is made to the received / reply text database 190 of FIG. For example, when “I'm busy tomorrow tomorrow” is selected as a reply message for the received sentence “Would you like to go to the movie tomorrow?”, The “reject” column 198 that is the attribute is set as the presentation order priority information. In the reply sentence candidate extraction process for the next partial character string sequence T (R), based on this presentation order priority information, the reply sentence candidates in the “reject” column 198 are preferentially presented at the top, and other replies You can present the sentence continuously. For example, if the next received sentence is "Can you lend me the previous CD?", The reply sentence "I'll lend you again" in the "Reject" column 198 is displayed at the top when presenting the reply sentence candidate. To do.

  Such display adjustment is based on the fact that, for example, in emails composed of a plurality of topics, there is often a connection between topics, and there is a high possibility of replying with the same meaning throughout. Done.

  The setting of the reply sentence candidate presentation order based on the attribute information is not limited to this. For example, an attribute for the user to reply in advance may be set, and an attribute to be selected according to the reply partner of the mail may be set. It may be set.

  According to such a configuration, the display order of the subsequent reply sentence candidates can be changed according to the reply sentence selected immediately before, and the reply sentence that is likely to be selected by the user can be displayed higher. become. The number of cursor operations performed by the user to select a reply sentence can be reduced, and the burden on the user can be further reduced.

  [Advantages, features and other embodiments of the first, second, third, fourth, fifth, sixth and seventh embodiments]

  (1) This information processing device is a received sentence / reply sentence database storage means for storing a received sentence and one or more reply sentences corresponding to the received sentence and a mail that divides the received mail into partial character strings by a predetermined method. Dividing means is provided. Search means for searching for a similar received sentence from the received sentence / reply sentence database stored in the received sentence / reply sentence database storage means for a series of one or more partial character strings divided by the mail dividing means. Reply sentence candidate extracting means for extracting one or more reply sentences set for the retrieved received sentence as reply sentence candidates is provided. Reply sentence candidate display means for prompting the user to select a reply sentence, a reply sentence candidate display means for prompting the user to select a reply sentence, a selection input means for inputting the user's selection, and a reply sentence combination that combines the selected reply sentences to generate a reply mail Means and the like.

  (2) The received text / reply text database sets received text and one or more reply texts in pairs, and indicates whether the device can automatically select a reply text for each set. Stores automatic reply availability information. The automatic selection / user selection switching unit refers to the automatic reply availability information for the reply sentence candidate and determines whether the apparatus automatically selects the reply sentence or presents the reply sentence candidate to the user to prompt the user to select a reply sentence. Switch. The reply sentence automatic selection unit automatically selects reply sentences for reply sentence candidates determined to be automatically selected. The reply sentence candidate display unit presents reply sentence candidates to the user for the reply sentence candidates determined to be selected by the user, and prompts the user to select a reply sentence. The reply sentence combining unit is configured to generate a reply mail by connecting the reply sentence automatically selected and the reply sentence selected by the user.

  (3) The reply message combining unit is configured to generate a reply mail by using the selected reply text and the corresponding partial character string of the received mail.

  (4) The specific information storage unit stores the date and location included in the mail. The received text / reply text database storage unit is a pair of the received text and one or more reply texts, and a character string representing the date / time and location included in the received text / reply text is the date / time Replace with the tag indicating and store. The specific information extraction unit. The date and place included in the partial character string are extracted based on the date and place detection rule. In addition, the specific information replacement processing unit extracts one or more reply sentences set for the received received sentence as reply sentence candidates, and the reply sentence includes a tag indicating the date and time and location Is configured to return to the date and time before the replacement in the received mail.

  (5) The expression replacement unit is configured to replace a specific character string included in the mail with another character string.

  (6) The received text / reply text database sets the received text and one or more reply texts as a set, and further stores attributes corresponding to the contents of each reply text. A reply part candidate display adjustment part is the structure which changes the presentation order of the reply sentence of the reply sentence candidate newly shown according to the attribute of the reply sentence which the user selected in the past.

  (7) The information processing device automatically presents reply sentence candidates to the user according to the contents of various sentences (received sentences) included in the received mail, and provides an interface for the user to select them. prepare. Thereby, a reply mail to an electronic mail written in a colloquial sentence including a plurality of topics can be generated with less burden on the user.

  (8) This information processing apparatus stores a combination of a received sentence and a reply sentence in a database in units of sentences. Divide the received mail into partial character strings of sentence length at appropriate positions, and receive similar data from the database in various lengths, from individual cases to consecutive sub character strings for each partial character string The sentence is searched, and the reply sentence candidate is displayed and selected by the user. By creating a reply mail by connecting the selected reply sentences, a reply mail for an electronic mail written in a spoken sentence containing a plurality of topics is generated.

  (9) In the above embodiment, the mobile phone 40 is shown as an example of an information processing apparatus that executes the function of creating a reply mail. For example, as an information processing apparatus having an electronic mail function, for example, as shown in FIG. 45, a personal digital assistant (PDA: Personal Digital Assistant) 300 may be used. The PDA 300 includes a display unit 302 and an operation input unit 304, for example. In addition, any information processing device having an electronic mail function may be used.

  (10) In the above embodiment, in each embodiment, an automatic reply function (FIG. 27), a reply mail display function (FIG. 30), a specific information replacement function (FIG. 33), an expression replacement function (FIG. 38), a reply Although the sentence display replacement function (FIG. 42) and the like have been described individually, the present invention is not limited to this. That is, a reply mail may be created by combining a plurality of these functions.

  Next, the following additional notes will be disclosed with respect to the embodiment including the above-described examples. The present invention is not limited to the following supplementary notes.

(Supplementary note 1) a division function unit that divides the received mail body into a plurality of partial character strings;
A storage unit that stores received sentence candidates corresponding to one or a plurality of consecutive partial character strings, and reply sentence information associated with the received sentence candidates;
A received sentence selection unit that selects, as a received sentence, the received sentence candidate that matches or is most similar to all of the one or a plurality of consecutive partial character strings;
A search range setting unit that sets a combination of the received texts that are the most similar to the mail text, and a search range setting unit that sets all of the divided partial character strings;
A reply sentence extraction unit that extracts reply sentence information corresponding to the received sentence stored in the storage unit based on the search range;
An information processing apparatus comprising:

(Supplementary Note 2) The received sentence selection unit calculates a similarity between one or a plurality of consecutive partial character strings and the received sentence candidate stored in the storage unit, and based on the calculated similarity, the A similarity calculation unit that selects the received sentence with respect to the partial character string;
The information processing apparatus according to appendix 1, wherein the search range setting unit selects the combination of the received sentences in which the partial character strings are continuous and the calculated cumulative value of the similarity is the maximum. .

(Additional remark 3) The said received sentence selection part selects the said received sentence with respect to all the one or several continuous said partial character strings according to the character string order of the said mail text,
The information processing apparatus according to appendix 1 or 2, wherein the search range setting unit combines the received texts in accordance with a character string order of the mail text.

(Additional remark 4) The said memory | storage part stores the automatic response information storage part which stores the automatic response availability information which makes it extract automatically the reply message information set with respect to each received sentence candidate,
The reply sentence extracting unit refers to the automatic response rating information corresponding to the reply sentence information extracted based on the set search range, and an automatic response switching unit that switches to an automatic response;
If the automatic response is set for the received sentence candidate, the reply sentence information in the storage unit is automatically extracted and displayed. apparatus.

(Supplementary Note 5) A display control unit that displays one or a plurality of continuous partial character strings set in the search range and the extracted reply text information in association with each other,
The information processing apparatus according to appendix 1 or 2, wherein at least one of the partial character string or the reply sentence information is displayed with an identification symbol identifying the partial character string or the reply sentence information. .

(Additional remark 6) The specific information extraction part which extracts the specific information contained in the said partial character string,
A specific information storage unit for storing the extracted specific information;
A specific information replacement processing unit that replaces the specific information included in the partial character string with a conversion tag, and replaces the conversion tag included in the reply sentence information extracted from the storage unit with the specific information;
The storage unit stores the reply sentence information in which the conversion tag is set in a part corresponding to the specific information,
The specific information replacement processing unit replaces the conversion tag with the specific information when the conversion tag is set in the extracted reply message information. Information processing device.

(Supplementary Note 7) A replacement character string storage unit that stores a specific character string in association with another character string having the same or similar meaning,
A replacement unit that replaces the specific character string included in the partial character string with the other character string in the replacement character string storage unit;
The information processing apparatus according to claim 1 or 2, wherein the received sentence selection unit selects the received sentence candidate based on the partial character string replaced with the other character string.

(Supplementary Note 8) The storage unit includes an attribute information storage unit that stores attribute information for classifying the reply sentence information.
A reply sentence candidate display adjustment unit that preferentially displays the reply sentence candidates classified into the same attribute based on the attribute information of the reply sentence selected in the past;
The information processing apparatus according to appendix 1 or 2, characterized by comprising:

(Additional remark 9) The said memory | storage part is equipped with the database which linked | related the received sentence candidate corresponding to the said partial character string of at least 1 or several continuous, and the said reply sentence information, Additional remarks 1, 2, The information processing apparatus according to 6, 7, or 8.

(Supplementary note 10) The information processing apparatus according to supplementary note 4, wherein the storage unit includes at least a database in which each received sentence candidate is associated with the automatic response availability information.

(Supplementary Note 11) The information processing apparatus according to Supplementary Note 7, wherein the replacement character string storage unit includes at least a database that associates the specific character string with the other character string.

(Additional remark 12) It is the reply sentence extraction method of the email of information processing apparatus,
Divide the received email body into multiple substrings,
Storing a received sentence candidate corresponding to one or a plurality of consecutive partial character strings and reply sentence information associated with the received sentence candidate in a storage unit;
The received sentence candidate that matches or is most similar to all of one or a plurality of consecutive partial character strings is selected as a received sentence,
All of the divided partial character strings are continuous, and the combination of the received sentence that is most similar to the mail body is set as a search range,
A reply sentence extraction method for an e-mail, wherein reply sentence information corresponding to the received sentence stored in the storage unit is extracted based on the search range.

(Supplementary note 13) Further, the similarity between one or a plurality of consecutive partial character strings and the received sentence candidate stored in the storage unit is calculated, and the partial character string is calculated based on the calculated similarity. Select the received text,
13. The method for extracting a reply message of an e-mail according to appendix 12, wherein the combination of the received texts in which the partial character strings are continuous and the calculated cumulative value of the similarity is the maximum is selected.

(Additional remark 14) Furthermore, according to the character string order of the said mail text, the said received sentence is selected with respect to all the one or several continuous said partial character strings,
14. The method of extracting an e-mail reply sentence according to appendix 12 or 13, wherein the received sentences are combined according to a character string order of the mail body.

(Supplementary Note 15) Further, automatic response availability information for automatically extracting reply text information set for each received text candidate is stored in the storage unit,
Switch to an automatic response with reference to the automatic response rating information corresponding to the reply sentence information extracted based on the set search range,
The reply message of the e-mail according to appendix 12 or 13, characterized in that, when an automatic response is set for the received sentence candidate, the reply sentence information in the storage unit is automatically extracted and displayed. Extraction method.

(Supplementary Note 16) One or a plurality of continuous partial character strings set in the search range and the extracted reply sentence information are displayed in correspondence with each other,
14. The e-mail according to appendix 12 or 13, wherein at least one of the partial character string or the reply sentence information is displayed with an identification symbol identifying the partial character string or the reply sentence information. Reply text extraction method.

(Supplementary Note 17) Extracting specific information included in the partial character string,
The extracted specific information is stored in a specific information storage unit,
Replacing the specific information included in the partial character string with a conversion tag, and replacing the conversion tag included in the reply sentence information extracted from the storage unit with the specific information,
Storing the reply sentence information in which the conversion tag is set in a part corresponding to the specific information in the storage unit;
14. The e-mail reply sentence extraction method according to appendix 12 or 13, wherein when the conversion tag is set in the extracted reply sentence information, the conversion tag is replaced with the specific information.

(Supplementary Note 18) A specific character string and another character string having the same or similar meaning are associated with each other and stored in the replacement character string storage unit,
Replacing the specific character string included in the partial character string with the other character string in the replacement character string storage unit;
14. The e-mail reply sentence extraction method according to appendix 12 or 13, wherein the received sentence candidate is selected based on the partial character string replaced with the other character string.

(Supplementary Note 19) Attribute information for classifying the reply sentence information is stored in the storage unit,
The reply message of the e-mail according to appendix 12 or 13, wherein the reply message candidates classified into the same attribute are preferentially displayed based on the attribute information of the reply message selected in the past. Extraction method.

(Supplementary note 20) An e-mail reply sentence extraction program to be executed by a computer of an information processing apparatus,
Divide the received email body into multiple substrings,
Storing a received sentence candidate corresponding to one or a plurality of consecutive partial character strings and reply sentence information associated with the received sentence candidate in a storage unit;
The received sentence candidate that matches or is most similar to all of one or a plurality of consecutive partial character strings is selected as a received sentence,
All of the divided partial character strings are continuous, and the combination of the received sentence that is most similar to the mail body is set as a search range,
An e-mail reply sentence extraction program that extracts reply sentence information corresponding to the received sentence stored in the storage unit based on the search range.

As described above, the information processing apparatus of the present disclosure, the e-mail reply sentence extraction method, and the preferred embodiment of the program thereof have been described. However, the present invention is not limited to the above description, It goes without saying that various modifications and changes can be made by those skilled in the art based on the gist of the invention described in the claims or disclosed in the embodiments for carrying out the invention. Needless to say, it is included in the scope of the present invention.

2 Information processing device 4, 131, 141, 151, 171, 191 Mail creation device 6 Received mail 8 Dividing function unit 10 Received sentence selection unit 12, 58 Storage unit 14 Search range setting unit 16 Reply sentence extracting unit 20 Mail dividing unit 22 Search unit 24, 130, 190 Received sentence / reply sentence database 26 Reply sentence combination part 28 Reply sentence candidate display part 30 Selection input part 40 Mobile phone 48 Operation input part 50 Display part 64 Extraction information buffer
66, 140 Display control section 80 Received text field 82 Reply text candidate field 84 Heading field 86 Data number field 88 Data number field 90 Reply text 92 Mail screen 106 Received mail area 108 Selection candidate area 132 Automatic reply availability information 134 Automatic selection / user Selection switching section 136 Reply sentence automatic selection section 138 Automatic reply availability column 142, 156 Reply sentence generation section 150 Specific information extraction section 152 Specific information storage section 154 Specific information replacement processing section 158, 164, 166 Tag 170 Expression replacement section 172 Expression replacement Character string table 192 Attribute information storage unit 193 Reply sentence candidate display adjustment unit

Claims (8)

A split function part that splits the received email body into multiple partial character strings;
A storage unit that stores received sentence candidates corresponding to one or a plurality of consecutive partial character strings, and reply sentence information associated with the received sentence candidates;
A received sentence selection unit that selects, as a received sentence, the received sentence candidate that matches or is most similar to all of the one or a plurality of consecutive partial character strings;
A search range setting unit that sets a combination of the received texts that are the most similar to the mail text, and a search range setting unit that sets all of the divided partial character strings;
A reply sentence extraction unit that extracts reply sentence information corresponding to the received sentence stored in the storage unit based on the search range;
An information processing apparatus comprising:
The storage unit stores an automatic response information storage unit that stores automatic response availability information that automatically extracts reply message information set for each received sentence candidate;
The reply sentence extracting unit refers to the automatic response rating information corresponding to the reply sentence information extracted based on the set search range, and an automatic response switching unit that switches to an automatic response;
2. The information processing apparatus according to claim 1, wherein, when an automatic response is set for the received sentence candidate, the reply sentence information in the storage unit is automatically extracted and displayed. .
A display control unit for displaying one or a plurality of continuous partial character strings set in the search range and the extracted reply sentence information in association with each other;
The information processing apparatus according to claim 1, wherein at least one of the partial character string or the reply sentence information is displayed with an identification symbol identifying the partial character string or the reply sentence information.
A specific information extraction unit that extracts specific information included in the partial character string;
A specific information storage unit for storing the extracted specific information;
A specific information replacement processing unit that replaces the specific information included in the partial character string with a conversion tag, and replaces the conversion tag included in the reply sentence information extracted from the storage unit with the specific information;
The storage unit stores the reply sentence information in which the conversion tag is set in a part corresponding to the specific information,
The information according to claim 1, wherein the specific information replacement processing unit replaces the conversion tag with the specific information when the conversion tag is set in the extracted reply sentence information. Processing equipment.
A replacement character string storage unit that stores a specific character string in association with another character string having the same or similar meaning;
A replacement unit that replaces the specific character string included in the partial character string with the other character string in the replacement character string storage unit;
The information processing apparatus according to claim 1, wherein the received sentence selection unit selects the received sentence candidate based on the partial character string replaced with the other character string.
The storage unit includes an attribute information storage unit that stores attribute information for classifying the reply sentence information.
A reply sentence candidate display adjustment unit that preferentially displays the reply sentence candidates classified into the same attribute based on the attribute information of the reply sentence selected in the past;
The information processing apparatus according to claim 1, further comprising:
An information processing apparatus e-mail reply sentence extraction method comprising:
Divide the received email body into multiple substrings,
Storing a received sentence candidate corresponding to one or a plurality of consecutive partial character strings and reply sentence information associated with the received sentence candidate in a storage unit;
The received sentence candidate that matches or is most similar to all of one or a plurality of consecutive partial character strings is selected as a received sentence,
All of the divided partial character strings are continuous, and the combination of the received sentence that is most similar to the mail body is set as a search range,
A reply sentence extraction method for an e-mail, wherein reply sentence information corresponding to the received sentence stored in the storage unit is extracted based on the search range.
An e-mail reply sentence extraction program to be executed by a computer of an information processing apparatus,
Divide the received email body into multiple substrings,
Storing a received sentence candidate corresponding to one or a plurality of consecutive partial character strings and reply sentence information associated with the received sentence candidate in a storage unit;
The received sentence candidate that matches or is most similar to all of one or a plurality of consecutive partial character strings is selected as a received sentence,
All of the divided partial character strings are continuous, and the combination of the received sentence that is most similar to the mail body is set as a search range,
An e-mail reply sentence extraction program that extracts reply sentence information corresponding to the received sentence stored in the storage unit based on the search range.

Images