JP2005157547A - Similar article extracting method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a new technology to present proper similar data. <P>SOLUTION: This similar article extracting method comprises a step for receiving message data as retrieval conditions, and for storing the message data in a message storing part, a step for extracting article data corresponding to the message data stored in the message data storing part from an article data storing part by using a predetermined algorithm, and for storing the message data in an extracted article data storing part, a step for extracting keyword candidates different from a keyword for retrieval preliminarily associated with the article data stored in the extracted article data storing part from among keyword candidates included in the message data stored in the message data storing part when predetermined conditions are satisfied, and for storing the keyword candidates in a non-solution keyword storing part and a step for associating the keyword candidates stored in the non-solution keyword storing part or words and phrases designated by a user as exclusion keywords with the article data stored in the extracted article data storing part, and for updating and registering the article data in the article data storing part. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a technique for searching for similar articles.

  In a system that accepts questions from users and presents answers, a technique that presents users with past data containing similar questions is generally used to save the trouble of individually answering all questions. It has been.

  For example, there is a technique for identifying a question paragraph and an answer paragraph by analyzing a paragraph in an e-mail and automatically answering a question similar to a past question (for example, Patent Document 1). reference). In other words, the paragraph function analysis means analyzes the paragraph of the question from the body of the e-mail based on the paragraph function decision rules of the paragraph function decision rule storage means. If the previous paragraph is a question and a quoted sentence, the paragraph of the answer is determined based on the paragraph function decision rule that the paragraph is an answer. When a question paragraph is included in the e-mail, the answer search means extracts an arbitrary keyword from the question paragraph, and based on this keyword, the e-mail has a question having a similar keyword and has the answer. An email is searched, and an email response obtained as a result of the search is determined as an answer to the email of the question.

  Further, for example, there is a technique in which a keyword is extracted from question data, a past similar question is searched based on the keyword, and the user is shown (for example, see Patent Document 2). That is, using a question management system in which a question sending terminal and a question receiving server are connected via a network, a new question inputted at the terminal is sent to the server, and a dictionary and past questions held in the storage device of the server The keyword is extracted from the new question while referring to, the past question in which one or more keywords match the keyword in the past question is transmitted to the question transmission terminal, and all of the transmitted past questions are different from the new question. When the instruction is received from the question transmission terminal, the new question is held in the storage device of the server.

In addition, for example, there is a technology that enables a user to search for similar questions in an educational system (see, for example, Patent Document 3). That is, when a student who is a questioner makes an answer request, he / she makes an inquiry based on the priority order for a plurality of answerers set in the answerer information, and transmits the question to the available answerers. Also, if all respondents are answering, send a request event to all respondents set in the respondent information, store the question in the question waiting buffer, wait until the current answer processing ends, If a respondent receives a signal to start a question based on the request event, ask the question. Also, if the request event time limit has passed, the student can obtain an answer later by sending the contents of the question waiting buffer as an e-mail to the respondent as it is, or past questions and their answers are accumulated. By searching for similar questions by referring to a database, if there is a similar question, an answer is obtained at that time.
Japanese Patent Laid-Open No. 10-25495 JP 2002-324163 A JP-A-8-235089

  However, according to such a conventional technique, for example, when a user posts a question, similar past question and answer data may be presented to the user in large quantities. The user must find out the data he needs from the large amount of data presented. However, since it is similar data to the last, there is not always what the user is seeking. For this reason, it is conceivable that it becomes difficult for the user to obtain an answer as the amount of accumulated data increases. For example, even if a certain user refers to similar data for a specific question and finds that there is no data required, for example, when another user posts a similar question, similar similar data is presented. Will be. That is, it cannot be said that similar data appropriate for the user can be referred to.

  In such a question answering system or the like, if a strict search is performed such as narrowing the range to be determined as similar, the number of cases where similar data is not searched increases, which is not appropriate.

  Therefore, an object of the present invention is to provide a novel technique for presenting appropriate similar data.

  In the method according to the present invention, when message data as a search condition is received, the message data stored in the message data storage unit is stored in the message data storage unit using a predetermined algorithm. The step of extracting corresponding article data from the article data storage unit and storing it in the extracted article data storage unit, and a keyword included in the message data stored in the message data storage unit when a predetermined condition is satisfied Of the candidates, extracting a keyword candidate different from the search keyword previously associated with the article data stored in the extracted article data storage unit, storing the keyword candidate in the unresolved keyword storage unit, Stored keyword candidates or phrases specified by the user as negative keywords And updating registration in the article data storage unit in association with the article stored in the output article data storage unit data.

  In this way, by associating an excluded keyword with article data to be searched, it is possible to perform a search using an excluded keyword in addition to a normal search keyword search. For example, it is possible to perform a search process in which article data associated with the same exclusion keyword is not extracted for a search condition (message data) including a specific exclusion keyword. According to this, article data shown to the user can be narrowed down.

  In addition, the predetermined algorithm includes extraction of first article data associated with search keywords that match at least some of the words included in the message data, and among the extracted first article data An algorithm for extracting second article data excluding article data associated with excluded keywords that match at least some of the phrases included in the message data may be used.

  Thereby, the 1st article data extracted by the search keyword and the 2nd article data which narrowed down the 1st article data by the exclusion keyword can be specified. For example, the first article data or the second article data can be indicated by a user request.

  Further, the user confirms the content of the first article data in accordance with a step of generating and outputting data for the user to confirm the content of the second article data, and a request from the user who confirmed the content of the second article data. A step of generating and outputting data for performing the processing.

  In this way, the second article data, which is data that has been narrowed down (highly likely to be small in number), is first shown to the user, and if there is a request from the user, the first article data that has not been narrowed down is shown. Thus, it is possible to cope with the case where narrowing down by the excluded keyword does not work effectively.

  The predetermined condition may be to detect an instruction from the user that the content of the article data stored in the extracted article data storage unit is insufficient. In this way, if an exclusion keyword is set for article data that has not been searched for by the user but hit the search, for example, even if a search is performed under the same search condition by another user thereafter. Article data in which an exclusion keyword is set (article data that is highly likely not to be an article requested by the user in the search) is not presented. That is, convenience for other users is enhanced.

  It is also possible to create a program for causing a computer to execute the method according to the present invention, and the program is a storage medium such as a flexible disk, a CD-ROM, a magneto-optical disk, a semiconductor memory, and a hard disk. Alternatively, it is stored in a storage device. Moreover, it may be distributed as a digital signal via a network. Note that data being processed is temporarily stored in a storage device such as a computer memory.

  ADVANTAGE OF THE INVENTION According to this invention, the similar data which excluded the thing which has a high possibility of being unnecessary for a user can be shown to a user, and suitable similar data can be shown.

  A system configuration diagram according to an embodiment of the present invention is shown in FIG. For example, in the network 1 which is an in-house network, for example, one or a plurality of user terminals 3 which are personal computers, a person-in-charge terminal 7 and an administrator terminal 9, and an electronic bulletin board service for posting or searching for questions and answers are provided. A QA (Question and Answer) server 5 to be performed is connected wirelessly or by wire. The network 1 may be another network such as the Internet.

  The user terminal 3, the person-in-charge terminal 7, the administrator terminal 9, and the QA server 5 are computer devices as shown in FIG. 2, and include a memory 201, a CPU 203, a hard disk drive (HDD) 205, and a display device 209. A display control unit 207 to be connected, a drive device 213 for the removable disk 211, an input device 215, and a communication control unit 217 for connecting to a network are connected by a bus 219. An application program including an operating system (OS) and a program for realizing processing in the present embodiment is stored in the HDD 205, and is read from the HDD 205 to the memory 201 when executed by the CPU 203. It is. If necessary, the CPU 203 controls the display control unit 207, the communication control unit 217, and the drive device 213 to perform necessary operations. Further, data in the middle of processing is stored in the memory 201, and if necessary, stored in the HDD 205. A program for realizing the processing in the embodiment of the present invention is stored in the removable disk 211 and distributed, for example, is received from the drive device 213 or via the network and communication control unit 217, and is installed in the HDD 205. Such a computer apparatus realizes various functions described below by organically cooperating hardware such as the CPU 203 and the memory 201 described above with the OS and necessary application programs.

  Returning to the description of FIG. 1, the user terminal 3 and the person-in-charge terminal 7 each have a web browser function, and receive and display web page data transmitted from the QA server 5 via the network 1, for example. It is possible to do. The administrator terminal 9 has a function of accessing each database (DB) described below, and enables registration or correction of data by a system administrator or the like.

  The QA server 5 includes a DB access unit 51, a part-of-speech decomposition unit 53, an unresolved keyword extraction unit 55, a Web server unit 57, and a work memory area 500. A QDB 560, a keyword DB 570, and a person-in-charge DB 580 are included. It is connected. The QDBB 560, the keyword DB 570, and the person-in-charge DB 580 are not necessarily configured as separate DBs. For example, a table corresponding to each of the above DBs may be included in one DB. Further, the work memory area 500 of the QA server 5 includes a posted data storage unit 510, a similar article data storage unit 520, a narrowed article data storage unit 530, a decomposition keyword storage unit 540, and an unresolved keyword storage unit 550. It is.

  A specific description of each processing unit and each storage unit will be described later, but the DB access unit 51 registers data in the QADB 560, the keyword DB 570, and the person-in-charge DB 580, or performs a search process for registered data. Do. The part-of-speech decomposition unit 53 performs part-of-speech decomposition on message data that is, for example, a question message, for example, to make data in units of words. Note that conventional techniques may be used for the method of part-of-speech decomposition, and detailed description thereof is omitted. Further, although the present embodiment is configured to perform Japanese part-of-speech decomposition, other parts of speech may be decomposed.

  The unresolved keyword extraction unit 55 extracts a keyword that is included in the question message that is the search condition and that is not included in the search keyword associated with the article data hit in the search. For example, when a user posts a question and refers to a similar article, the answer that the user requests may not be included in the similar article. As will be described later, a search keyword is associated with each article, and an article in which the search keyword is included in the question message is presented as a similar article. If the answer requested by the user is not included in a similar article, there is usually a keyword that is included in the question message and not included in the search keywords for similar articles. Extract as keywords.

  The Web server unit 57 has a general Web server function and generates, for example, Web page data and transmits it to the user terminal 3 or the like via the network 1. The post data storage unit 510 is a storage unit for temporarily storing post data from the user such as questions and answers. The similar article data storage unit 520 is a storage unit for temporarily storing article data hit in the similar article search. The narrowed article data storage unit 530 is a storage unit for temporarily storing article data that has been hit by a similar article search and has been narrowed down by an excluded keyword.

  The decomposition keyword storage unit 540 is a storage unit for temporarily storing the message data decomposed, for example, in units of words by the part of speech decomposition unit 53, that is, the processing result of the part of speech decomposition unit 53. The unresolved keyword storage unit 550 temporarily stores a keyword that does not hit the search keyword among the data stored in the decomposition keyword storage unit 540, that is, a processing result of the unresolved keyword extraction unit 55. Part.

  FIG. 3 shows an example of the table configuration of QADB 560 and stored data. 3 includes an article number (NO) column 300, a type column 302, a questioner email address column 304, a questioner name column 306, a questioner department column 308, a question date column 310, and a question. Column 312, title column 314, respondent name column 316, respondent department column 318, answer date / time column 320, and answer column 322. The record is generated when a new question is received from a user who is a questioner, and includes an article number (NO) column 300, a type column 302, a questioner email address column 304, and a questioner name column 306. Values are registered in the questioner department column 308, the question date and time column 310, the question column 312, and the title column 314. Here, the new question refers to a question that cannot be answered even if a similar article is referenced. Then, when accepting a reply from the respondent, values are registered in the respondent name column 316, the respondent department column 318, the reply date / time column 320, and the reply column 322.

  The record is uniquely identified by the value in the column 300 of the article number (NO). Further, values are automatically registered in the article number (NO) column 300, the question date column 310, and the answer date column 318 by the system. In the type column 302, the questioner determines which category the question is classified into, and the selected item is registered. In addition, a predetermined code may be registered in the type column 302, the questioner name column 306, the questioner department column 308, the respondent name column 314, and the answerer department column 316. .

  FIG. 4 shows an example of the table configuration of the keyword DB 570 and stored data. The example of FIG. 4 includes a corresponding article number (NO) column 400, a similar article search keyword column 402, and a keyword column 404 that has not been resolved. Records are generated corresponding to the records of the QADB 560, and values are first registered in the column 400 of the corresponding article number (NO) and the column 402 of similar article search keywords. In the corresponding article number (NO) column 400, the value of the article number (NO) column 300 of the corresponding QADB 560 record is registered, and the record is uniquely identified by the value of the corresponding article number (NO) column 400. The Then, in the keyword column 404 that has not been solved, the excluded keywords described above are stored. For example, “homepage address” is registered in the records from the first line to the third line in the example shown in FIG. 4, and this is because the article data corresponding to the record includes the keyword “homepage address” ( (Question) indicates that no answer is given. Since the QADB 560 table and the keyword DB 570 table have the same item content as the primary key, they may be combined into one table.

  FIG. 5 shows an example of the table configuration of the person-in-charge DB 580 and stored data. The example of FIG. 5 includes a type column 500, a person in charge / department signature column 502, and a mail address column 504. Records are set and registered when the system administrator operates the administrator terminal 9, for example. Further, the record is uniquely identified by the value in the type column 500. In the person-in-charge / department-signature column 502, the name of the person in charge of the answer or the department corresponding to the question type (value in the type column 500) is registered. That is, for example, the system administrator sets such that “this person answers the question of this classification”. When the department in charge is set instead of the person in charge, for example, the mailing list address of the department is registered in the mail address column 504 instead of the personal mail address.

  Next, processing contents of the system shown in FIG. 1 will be described with reference to FIGS. As a premise of the processing flow shown in FIG. 6, a question input page is displayed on the display screen of the user terminal 3 by processing of the Web browser of the user terminal 3 and the Web server unit 57 of the QA server 5. It shall be. If login is required when using the system, login processing is also performed. First, a Web browser of the user terminal 3 (hereinafter abbreviated as “user terminal 3”) receives a question input and a posting request from a user who is a questioner (FIG. 6: step S1). Although not shown, input of question data and a posting request are accepted by, for example, a text box or a posting button included in the displayed page. Then, the user terminal 3 transmits the input question data to the QA server 5 (step S3).

  The Web server unit 57 of the QA server 5 receives the question data from the user terminal 3 and stores it in the posted data storage unit 510 of the work memory area 500 (step S5). The received question data includes the type of question, the email address of the questioner, the name of the questioner, the department of the questioner, the title, and the question text (not necessarily a sentence), and is the date and time of reception. Store as question data including question date and time. The questioner's email address, the name of the questioner, and the department of the questioner do not necessarily need to be included in the question data at this point. When registering a new record in the QADB 560, The user may be prompted to input.

  Then, the DB access unit 51 of the QA server 5 refers to the keyword DB 570 and the QADB 560, and article data in which search keywords that match at least some of the phrases included in the question message (title and question sentence) are associated. (QADB 560 record) is extracted and stored in the similar article data storage unit 520 of the work memory area 500 (step S7).

  For example, if the question message is only the title and “Please tell me how to log in.” And the data shown in FIG. 4 is stored in the keyword DB 570, the search from the keyword DB 570 is “Login”. The records in the first and fourth lines with which the keywords (values in the similar article search keyword column 402) are associated are specified. Then, the QADB 560 is searched using “135” and “138” which are the values in the column 400 of the corresponding article number (NO) of the records in the first and fourth rows of the keyword DB 570. When the data as shown in FIG. 3 is stored in the QADB 560, the first row and the fourth row in which the value of the column 300 (FIG. 3) of the article number (NO) is “135” and the record of “138” are shown. The record of the line is extracted.

  Then, the DB access unit 51 determines whether a record is extracted in step S7 (step S9). If it is determined that the record has not been extracted (step S9: No route), the part of speech decomposition unit 53 of the QA server 5 performs part of speech decomposition on the question message to extract one or a plurality of keywords, It stores in the decomposition keyword storage unit 540 (step S11). If the question message is an enumeration of words, the part of speech may not be decomposed. Then, the DB access unit 51 registers a new record in the QADB 560 and the keyword DB 570 based on the question data and the keyword extracted in Step S11 (Step S13). And it transfers to the process of step S27 mentioned later.

  On the other hand, when it is determined in step S7 that a record has been extracted (step S9: Yes route), the DB access unit 51 of the QA server 5 refers to the keyword DB 570 and the QADB 560, and is extracted in step S7 and has a work memory area. Article data (excluding article data (similar article data) stored in 500 similar article data storage unit 520 except for those associated with excluded keywords that match at least some of the phrases included in the question message ( (Restricted article data) is extracted and stored in the narrowed article data storage unit 530 of the work memory area 500 (step S15).

  Specifically, first, the article number of the article data stored in the similar article data storage unit 520 is extracted. Then, referring to the keyword DB 570, the value (excluded keyword) of the unresolved keyword column 404 of the record corresponding to the extracted article number (the value of the corresponding article number (NO) column 400 matches) is Identify records that do not match at least some of the words in the message. Then, based on the value in the column 400 of the corresponding article number (NO) of the specified record, the QADB 560 (or similar article data storage unit 520) is searched, and the values in the column 300 of the corresponding (article number (NO)) match. Yes) to extract records (article data).

  Then, the Web server unit 57 generates QA list page data based on the narrowed-down article data and transmits it to the user terminal 3 (step S17). The user terminal 3 receives the QA list page data from the QA server 5 and displays it on the display device (step S19).

  FIG. 7 shows an example (part 1) of the QA list page. The example of FIG. 7 includes a question content display section 700, a “Yes” button 702, a “No” button 704, similar article display sections 706 to 710, and a “display all similar articles” button 712. In the example of FIG. 7, for example, in the example of the keyword DB 570 shown in FIG. 4, records in the first to third rows are registered, and nothing is registered in the keyword column 404 that has not been solved, and In the example of the QADB 560 shown in FIG. 3, the case where the records in the first to third rows are registered is shown. In such a case, when a question message as shown in the question content display section 700 is received, the search message registered in the similar article search keyword column 402 (FIG. 4) is included in the question message. Article data, that is, article data of article numbers (NO) 135 to 137 is extracted. Each article data is displayed in similar article display sections 706 to 710.

  In the example of FIG. 7, answers to questions from various users are shown, but an administrator or the like can register questions and answers that are assumed in advance. Then, the user looks at the data shown in the similar article display sections 706 to 710 and clicks the “Yes” button 702 when the question is solved, and clicks the “No” button 704 when the question is not solved. click. In the present embodiment, even if no answer is shown, if there is an article with the same question and if the answer is obtained, the question is solved, the “Yes” button 702 is clicked. Click. If all similar articles that have not been narrowed down are to be browsed, the “display all similar articles” button 712 is clicked. In addition, since the example of FIG. 7 is a case where no exclusion keyword is registered, the article presented is not changed even when the page is shifted. In such a case, the “display all similar articles” button 712 may not be included in the page.

  FIG. 8 shows an example (part 2) of the QA list page. The page structure is the same as that shown in FIG. 7, and includes a question content display section 800, a “Yes” button 702, a “No” button 704, a similar article display section 810, and a “Display all similar articles” button 712. It is. The example of FIG. 8 shows the case where the record as shown in FIG. 4 is registered in the keyword DB 570 and the case where the record as shown in FIG. 3 is registered in the QADB 560. In such a case, when a question message as shown in the question content display unit 800 is accepted, the search message registered in the similar article search keyword column 402 (FIG. 4) is included in the question message. Article data, that is, article data of article numbers (NO) 135 to 138 is extracted. Further, among these article data, the article data in which the excluded keyword registered in the unresolved keyword column 404 (FIG. 4) is not included in the question message, that is, the article data of the article number (NO) 138 is narrowed down. Extracted. Then, the extracted article data is displayed in the similar article display unit 810. In this way, when the excluded keyword is registered, article data that is assumed not to solve the accepted question is not presented.

  Returning to the description of FIG. 6, the user terminal 3 accepts a selection input from the user for the QA list page, and transmits the selection input data to the QA server 5 (FIG. 6: Step S21). The Web server unit 57 of the QA server 5 receives the selected input data from the user terminal 3 and temporarily stores it in the storage device (step S23). Then, the Web server unit 57 determines whether or not the selection input from the user is a selection input indicating “solved” (step S25). As described above, if the “yes” button 702 is clicked, it is determined that “solved”.

  If it is not determined that the problem has been resolved (step S25: No route), the process proceeds to the process of FIG. On the other hand, if it is determined that the problem has been resolved (step S25: Yes route), the web server unit 57 generates top page data of the QA system and transmits it to the user terminal 3 (step S27). The user terminal 3 receives the top page data from the QA server and displays it on the display device (step S29). Although not shown, the top page shows a menu of the QA system including a link button to a question input page. Then, the question acceptance process ends. As described above, similar articles extracted using the search keyword are narrowed down using the excluded keyword, and an appropriate QA list page is presented to the user.

  Next, processing after the transition through the terminal A will be described with reference to FIGS. 9 and 10. First, the Web server unit 57 of the QA server 5 determines whether the selection input from the user for the QA list page is “display all similar articles” (FIG. 9: step S41). As described above, if the “display all similar articles” button 712 is clicked, it is determined that “all similar articles are displayed”. If it is determined that “all similar articles are displayed” (step S41: Yes route), the process proceeds to the process of FIG. On the other hand, if it is determined that it is not “display all similar articles” (the “No” button 704 has been clicked) (step S41: No route), the part-of-speech decomposition unit 53 of the QA server 5 performs the work- The question message included in the question data stored in the posted data storage unit 510 of the memory area 500 is decomposed into parts of speech to extract keywords, and stored in the decomposition keyword storage unit 540 of the work memory area 500 (step S43).

  For example, if the title of the question message is “*** tyBANK login password” and the question is “I don't know the login ID and password. I forgot the homepage address. For example, keywords “*** tyBANK”, “login”, “password”, “ID”, and “homepage address” are extracted. The part-of-speech decomposition method and the keyword extraction method (for example, “I don't know” and “Tell me” are not extracted as keywords) may use conventional techniques and will not be described in detail.

  Then, the unresolved keyword extraction unit 55 of the QA server 5 extracts keywords that are not associated with any narrowed-down article data from the keywords stored in the decomposition keyword storage unit 540, and stores them in the work memory area 500. It stores in the unresolved keyword storage part 550 (step S45). For example, as shown above, the keywords “*** tyBANK”, “login”, “password”, “ID”, and “homepage address” are stored in the decomposition keyword storage unit 540 and associated with the narrowed-down article data. If the search keywords are “*** tyBANK”, “login”, and “password” (see, for example, the records in the first to third lines in FIG. 4), the keywords “ID” and “homepage address” are Extracted.

  Then, the Web server unit 57 generates “unresolved keyword” confirmation page data using the keywords stored in the unresolved keyword storage unit 550, and transmits the page data to the user terminal 3 (step S47). The user terminal 3 receives the “unresolved keyword” confirmation page data from the QA server 5 and displays it on the display device (step S49).

  FIG. 10 shows an example of the “unresolved keyword” confirmation page. The example of FIG. 10 includes an unresolved keyword display input field 1000, a “Yes” button 1002, a “No” button 1004, and a question content display unit 1006. In the unresolved keyword display input field 1000, keywords stored in the unresolved keyword storage unit 550 of the work memory area 500 in step S45 (FIG. 9) are shown. The unresolved keyword display input field 1000 allows the user to add, update, and delete keywords. For example, when the QA list page shown in FIG. 7 is transferred to the unresolved keyword confirmation page, “ID” among the keywords “ID” and “Homepage address” shown in the unresolved keyword display input field 1000 is displayed. Can be said to be deleted because it is shown in the answer portion of the similar article display portion 706 (FIG. 7). When the user can set the keyword (including the case where he / she resets the keyword) shown in the unresolved keyword display input field 1000 as an excluded keyword to be used in the subsequent search processing. A “Yes” button 1002 is clicked, and a “No” button 1004 is clicked when setting is not desired.

  Returning to the description of FIG. 9, the user terminal 3 accepts an input from the user for the “unresolved keyword” confirmation page, and the input data (excluded in the unresolved keyword display input field 1000 (FIG. 10)) (Including keywords) is transmitted to the QA server 5 (FIG. 9: step 51). The Web server unit 57 of the QA server 5 receives the input data from the user terminal 3 and temporarily stores it in the storage device (step S53). Then, the Web server unit 57 determines whether the instruction from the user is an instruction to “register” the excluded keyword (step S55). As described above, if the “Yes” button 1002 is clicked, it is determined that “Register” is selected.

  If it is determined that it is not “register” (the “No” button 1004 has been clicked) (step S55: No route), the process proceeds to step S59 described later. On the other hand, when it is determined that “register” (step S55: Yes route), the DB access unit 51 of the QA server 5 sets and registers the excluded keyword in the keyword DB 570 (step S57). For example, the excluded keyword shown in the unresolved keyword display input field 1000 (FIG. 10) is “homepage address”, and the article numbers (NO) of article data shown on the QA list page are 135 to 137. (See FIG. 7), as shown in the records in the first to third lines in the example of FIG. 4, the corresponding records in the keyword DB 570 (corresponding article numbers (NO) 135 to 137) were not resolved. A value (excluded keyword) “homepage address” is set and registered in the keyword column 404.

  Further, the DB access unit 51 registers a new record in the QADB 560 and the keyword DB 570 using the question data stored in the posted data storage unit 510 and the keyword stored in the decomposition keyword storage unit 540 (step S59). . And a process transfers to the process of step S27 (FIG. 6) via the terminal C. FIG.

  Next, a process after the transition through the terminal B will be described with reference to FIG. First, the Web server unit 57 of the QA server 5 generates QA list page data based on the similar article data stored in the similar article data storage unit 520 of the work memory area 500 and transmits it to the user terminal 3. (FIG. 11: Step S71). The user terminal 3 receives the QA list page data from the QA server 5 and displays it on the display device (step S73). The QA list page displayed here has the same configuration as the example shown in FIGS. However, since the article data that has not been narrowed down is shown, the page does not include the “display all similar articles” button 712.

  And the user terminal 3 receives the selection input from the user with respect to a QA list page, and transmits selection input data to the QA server 5 (step S75). The Web server unit 57 of the QA server 5 receives the selected input data from the user terminal 3 and temporarily stores it in the storage device (step S77). Then, the Web server unit 57 determines whether or not the selection input from the user is a selection input indicating “Resolved” (Step S79). If the “yes” button 702 is clicked, it is determined that “solved”. If it is determined that the problem has been resolved (step S79: Yes route), the process proceeds to the process of step S27 (FIG. 6) via the terminal C. On the other hand, if it is not determined as “solved” (step S79: No route), the part of speech decomposition unit 53 of the QA server 5 is included in the question data stored in the posted data storage unit 510 of the work memory area 500. The question message is decomposed to extract one or more keywords and stored in the decomposition keyword storage unit 540 of the work memory area 500 (step S81). If the question message is an enumeration of words, the part of speech may not be decomposed.

  Then, the unresolved keyword extraction unit 55 of the QA server 5 extracts keywords that are not associated with any similar article data from the keywords stored in the decomposition keyword storage unit 540, and stores them in the work memory area 500. It stores in the unresolved keyword storage part 550 (step S83). This process is the same as the process in step S45 (FIG. 9) described above, but here, similar article data is referred to instead of narrowed article data. And a process transfers to the process of step S47 via the terminal D. FIG.

  As described above with reference to FIGS. 9 to 11, the exclusion keyword is associated with the article data shown on the QA list page.

  In this way, appropriate similar article data is presented by using the excluded keyword, and setting and registration of the excluded keyword are performed.

  In the Web server unit 57 having the mail data generation and transmission function, a new record (question data) is registered in the QADB 560 (and the keyword DB 570) by the process of step S13 (FIG. 6) or step S59 (FIG. 9). When this is detected, the person in charge or the department in charge associated with the question type is identified with reference to the person in charge DB 580, and the URL (Uniform Resource Locator) of the answer page is addressed to the identified mail address. Send response request mail data including). The person in charge or the member of the department in charge who sees the reply request mail operates the person-in-charge terminal 7, for example, accesses the QA server 5 based on the URL of the answer page, and inputs the answer to the answer page. The input answer is set and registered in the QADB 560 by the DB access unit 51 of the QA server 5.

  Although one embodiment of the present invention has been described above, the present invention is not limited to this. For example, the table configurations shown in FIGS. 3 to 5 are merely examples, and other configurations may be adopted for storing similar data, and items are added or deleted as necessary. May be. The functional block configuration of the QA server 5 shown in FIG. 1 is merely an example, and may differ from the actual program / module configuration. Each device may be configured by a plurality of servers and computers. The functional block diagram of the computer shown in FIG. 2 is also an example, and may be different from the actual hardware configuration. In addition, the screen examples shown in FIGS. 7, 8, and 10 are examples, and the same data may be expressed in different modes. Furthermore, the processing flows shown in FIGS. 6, 9, and 11 are also examples, and the processing order may be changed within a range in which similar processing results can be obtained, and steps may be added or deleted as necessary. May be.

  Although the embodiment has been described by using an example applied to the QA system, it is not always necessary to handle article data composed of questions and answers. For example, searching for similar articles on an electronic bulletin board or enumeration of keywords (AND It is also possible to apply the technology according to the present invention to various other search systems, such as a homepage search according to (conditions).

(Appendix 1)
When message data as a search condition is received, storing in the message data storage unit;
Article data extraction step for extracting article data corresponding to the message data stored in the message data storage unit from the article data storage unit using a predetermined algorithm and storing it in the extracted article data storage unit;
When a predetermined condition is satisfied, among the keyword candidates included in the message data stored in the message data storage unit, the keyword data is associated with the article data stored in the extracted article data storage unit in advance. Extracting a keyword candidate different from the search keyword and storing it in the unresolved keyword storage unit;
The keyword candidate or the phrase specified by the user stored in the unresolved keyword storage unit is associated with the article data stored in the extracted article data storage unit as an excluded keyword and updated and registered in the article data storage unit And steps to
And a similar article extraction method executed by a computer.

(Appendix 2)
Among the extracted first article data, the predetermined algorithm includes extraction of first article data associated with search keywords that match at least some of the phrases included in the message data, and The algorithm for extracting second article data excluding article data associated with an exclusion keyword that matches at least a part of words included in the message data, according to claim 1 Similar article extraction method.

(Appendix 3)
Generating and outputting data for the user to confirm the content of the second article data;
Generating and outputting data for the user to confirm the content of the first article data in accordance with a request from the user who has confirmed the content of the second article data;
The similar article extraction method according to supplementary note 2, further including:

(Appendix 4)
Any one of Supplementary notes 1 to 3, wherein the predetermined condition is to detect an instruction from a user that the content of the article data stored in the extracted article data storage unit is insufficient. The similar article extraction method described in 1.

(Appendix 5)
A step of discarding the message data stored in the message data storage unit when detecting an instruction from the user that the content of the article data stored in the extracted article data storage unit is sufficient; The similar article extraction method according to any one of appendices 1 to 4.

(Appendix 6)
When the article data extraction step does not include article data corresponding to the message data, the message data and a keyword included in the message data are associated with the new article data and the new article data. The similar article extraction method according to any one of appendices 1 to 5, including a step of storing the retrieved keyword as a search keyword in the article data storage unit.

(Appendix 7)
Generating and outputting data for prompting a user to confirm the keyword candidate and specify a phrase stored in the unresolved keyword storage unit;
The similar article extraction method according to any one of supplementary notes 1 to 6, further including:

(Appendix 8)
When message data as a search condition is received, storing in the message data storage unit;
Using a predetermined algorithm, extracting the article data corresponding to the message data stored in the message data storage unit from the article data storage unit, and storing it in the extracted article data storage unit;
When a predetermined condition is satisfied, among the keyword candidates included in the message data stored in the message data storage unit, the keyword data is associated with the article data stored in the extracted article data storage unit in advance. Extracting a keyword candidate different from the search keyword and storing it in the unresolved keyword storage unit;
The keyword candidate or the phrase specified by the user stored in the unresolved keyword storage unit is associated with the article data stored in the extracted article data storage unit as an excluded keyword and updated and registered in the article data storage unit And steps to
A similar article extraction program that runs a computer.

(Appendix 9)
When message data as a search condition is received, means for storing in the message data storage unit;
Means for extracting the article data corresponding to the message data stored in the message data storage unit from the article data storage unit using a predetermined algorithm and storing it in the extracted article data storage unit;
When a predetermined condition is satisfied, among the keyword candidates included in the message data stored in the message data storage unit, the keyword data is associated with the article data stored in the extracted article data storage unit in advance. Means for extracting keyword candidates different from the search keyword and storing them in the unresolved keyword storage unit
The keyword candidate or the phrase specified by the user stored in the unresolved keyword storage unit is associated with the article data stored in the extracted article data storage unit as an excluded keyword and updated and registered in the article data storage unit Means to
A similar article extraction device.

1 is a system configuration diagram according to an embodiment of the present invention. It is a figure which shows the outline | summary of the functional block of the computer in embodiment of this invention. It is a figure which shows an example of a QADB table. It is a figure which shows an example of a keyword DB table. It is a figure which shows an example of a person in charge DB table. It is a figure which shows the processing flow (the 1) in one embodiment of this invention. It is a figure which shows an example (the 1) of a QA list page. It is a figure which shows an example (the 2) of a QA list page. It is a figure which shows the processing flow (the 2) in one embodiment of this invention. It is a figure which shows an example of an unresolved keyword confirmation page. It is a figure which shows the processing flow (the 3) in one embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Network 3 User terminal 5 QA server 7 Person in charge terminal 9 Manager terminal 51 DB access part 53 Part of speech decomposition part 55 Unresolved keyword extraction part 57 Web server part 500 Work memory area 510 Posted data storage part 520 Similar article data storage part 530 Narrowed article data storage unit 540 Decomposition keyword storage unit 550 Unsolved keyword storage unit 560 QADB 570 Keyword DB
580 Person in charge DB

Claims (5)

When message data as a search condition is received, storing in the message data storage unit;
Using a predetermined algorithm, extracting the article data corresponding to the message data stored in the message data storage unit from the article data storage unit, and storing it in the extracted article data storage unit;
When a predetermined condition is satisfied, among the keyword candidates included in the message data stored in the message data storage unit, the keyword data is associated with the article data stored in the extracted article data storage unit in advance. Extracting a keyword candidate different from the search keyword and storing it in the unresolved keyword storage unit;
The keyword candidate or the phrase specified by the user stored in the unresolved keyword storage unit is associated with the article data stored in the extracted article data storage unit as an excluded keyword and updated and registered in the article data storage unit And steps to
And a similar article extraction method executed by a computer.   Among the extracted first article data, the predetermined algorithm includes extraction of first article data associated with search keywords that match at least some of the phrases included in the message data, and 2. The algorithm for extracting second article data excluding article data associated with excluded keywords that match at least some of the words included in the message data. Similar article extraction method. Generating and outputting data for the user to confirm the content of the second article data;
Generating and outputting data for the user to confirm the content of the first article data in accordance with a request from the user who has confirmed the content of the second article data;
The similar article extraction method according to claim 2, further comprising:   4. The predetermined condition is to detect an instruction from a user that the content of the article data stored in the extracted article data storage unit is insufficient. The similar article extraction method according to one. When message data as a search condition is received, storing in the message data storage unit;
Using a predetermined algorithm, extracting the article data corresponding to the message data stored in the message data storage unit from the article data storage unit, and storing it in the extracted article data storage unit;
When a predetermined condition is satisfied, among the keyword candidates included in the message data stored in the message data storage unit, the keyword data is associated with the article data stored in the extracted article data storage unit in advance. Extracting a keyword candidate different from the search keyword and storing it in the unresolved keyword storage unit;
The keyword candidate or the phrase specified by the user stored in the unresolved keyword storage unit is associated with the article data stored in the extracted article data storage unit as an excluded keyword and updated and registered in the article data storage unit And steps to
A similar article extraction program that runs a computer.

Images