JP2008021077A - Case reference system, case reference method, case storage server, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an optimum case to be referred to when a person in charge of user support answers an inquiry from a user. <P>SOLUTION: A related case reference system 1 is provided with a terminal 10 in charge which receives inquiry mail from a user and returns an answer to the inquiry, a user terminal 30 for making the user send the inquiry mail in which an error message or an error log is described about a trouble concerned with an information processing system to the terminal 10 in charge and for receiving an answer, a case storage server 40 including a database 42 for storing cases of pieces inquiry mail and answer mail so as to retrieve each mail by keyword extraction, and a network 20. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a case reference system, a case reference method, and a program that present cases to be referred to when receiving and answering questions from users regarding troubles related to products, products, etc. The present invention relates to a case reference system, a case reference method, a case storage server, and a program that present cases to be referred to when answering a question on an error log.

  There is a task at a user support center or the like, where a person in charge of user support receives an inquiry about a trouble from a user and answers the user about the cause of the trouble and a countermeasure. In such a situation, it has been conventionally performed for a person in charge to obtain a case related to the question and to refer to and answer it. This is because not only knowledge and judgment of the person in charge but also answering with reference to an answer example for a similar question in the past will answer the question accurately and appropriately.

  Conventionally, an information processing technique for searching a case to be referred to when answering such a question has been studied. Specifically, an appropriate keyword is extracted from a question sentence to search for a case, and a case having the keyword is searched from a database. In this search technique, extracting an appropriate keyword from a character string included in a question sentence improves the quality and accuracy of search results.

  Therefore, there is a technique for improving the quality and accuracy of search results by automating the process of extracting keywords from sentences using knowledge and rules.

  In the technique described in Patent Document 1, a specific description part is detected by knowledge and rules from a sentence describing a problem in advance, and a keyword is extracted from the specific description part to create a case database as an index of the problem. A keyword is extracted from the question input by the user by the same method, and a sentence with high similarity is obtained by collating with the index stored in the case database.

  The technology described in Patent Document 2 obtains a search sentence that is a part of the natural sentence by obtaining a search sentence that is a part of the natural sentence by using a morphological analysis, a syntax analysis, and a sentence analysis of the natural sentence describing the problem using knowledge and rules. A similar case is searched by obtaining a similarity to the case sentence.

JP 2002-366556 A JP 2000-276487 A

  In the conventional technology described above, knowledge and rules are used to extract keywords from sentences to improve the quality and accuracy of search results, but knowledge and rules corresponding to sentences in various technical fields are set. It is extremely difficult to do. In addition, knowledge and rules must be maintained in order to keep up with technological advances, but this is not an easy task because it requires a high degree of expertise and a great number of man-hours. Therefore, there is a problem that even if the system according to the above-described conventional technology is introduced to the work of the user support center, the accuracy of the search result of the answer example is not sufficient and the support service that satisfies the user cannot be provided. .

  An object of the present invention is to provide a case reference system, a case reference method, a case storage server, and a program capable of presenting an optimal case to be referred to when a person in charge of user support answers a question from a user. It is to provide.

  The first case reference system of the present invention inputs question data, extracts a one-byte character string from the question data as a keyword, and searches the case database for case information corresponding to the keyword. And a case search means for reading out the information.

  The second case reference system of the present invention is the first case reference system, wherein the case information read by the case search means is output, and the answer processing means for inputting the answer data corresponding to the case information; Database management means for associating the answer data input by the answer processing means with the question data and storing the answer data as the case information in the case database.

  According to a third case reference system of the present invention, in the first or second case reference system, the keyword extraction unit is configured to perform predetermined processing on a plurality of one-byte character string word strings including spaces between words from the question data. A word string having a number of characters or more is extracted and used as the keyword.

  The first case reference method of the present invention includes a keyword extraction step in which a case storage server inputs question data, extracts a 1-byte character string from the question data as a keyword, and the case storage server includes the keyword And a case search step for searching and reading case information corresponding to the case data from the case database.

  The second case reference method of the present invention is the first case reference method, wherein the terminal device outputs the case information read in the case search step and inputs the answer data corresponding to the case information. An answer processing step, and the case accumulation server stores the answer data input in the answer processing step in the case database in association with the question data and stored as the case information in the case database. Features.

  According to a third case reference method of the present invention, in the first or second case reference method, the keyword extraction step is performed by using a predetermined word sequence of a plurality of 1-byte character strings including spaces between words from the question data. A word string having a number of characters or more is extracted and used as the keyword.

  The first case reference server of the present invention receives the question data transmitted from the terminal device, extracts a one-byte character string from the question data and uses it as a keyword, and case information corresponding to the keyword The case search means for reading out from the case database and transmitting it to the terminal device, the answer data corresponding to the case information from the terminal device is received, and the case data is associated with the question data as the case information. Database management means for storing in the database.

  According to a second case accumulation server of the present invention, in the first case reference server, the keyword extraction means has a predetermined number of characters or more for a plurality of one-byte character strings including a space between words from the question data. A word string is extracted and used as the keyword.

  A first program of the present invention receives a question data transmitted from a terminal device, extracts a one-byte character string from the question data as a keyword, and uses case information corresponding to the keyword as a case Case retrieval processing that reads from the database and transmits to the terminal device, receives answer data corresponding to the case information from the terminal device, and associates the answer data with the question data in the case database as the case information The database management processing to be stored is performed by a computer.

According to a second program of the present invention, in the first program, the keyword extraction process extracts a word string having a predetermined number of characters or more from the question data for a plurality of one-byte character strings including spaces between words. The keyword is used.
[Action]
The present invention aims to prevent human intervention and knowledge from intervening in information storage and retrieval. For this purpose, the following method is used. It should be noted that among the words in the question text, characters suitable for use in the search are concentrated on single-byte characters such as error messages, error logs, and device model numbers. In other words, if you search for a case (sentence) that contains the same error message or error log, or if you find a case (sentence) that contains the same model, the same program name or command name, and read it, you can find a hint This makes it easier to find out where things are going to be saved and where the reference materials are stored. This is not a simple full-text search, but utilizes the fact that common keywords are biased to single-byte characters that are easy for computers to handle. When one byte character is actually extracted from the sentence and the number is counted, if the number of words is small, the number of cases searched explosively increases and does not function as a keyword. However, if the keyword exceeds a certain number of characters, the number of searches suddenly decreases. An example of the verification will be described with reference to FIG. FIG. 15 shows a total of the number of searches performed by searching for 160,000 e-mails transmitted to the user support center with various keyword lengths using the method of the present invention. It is shown that the number of searches decreases when the keyword length is around 16 characters. In the embodiment described below, 16 characters are adopted as the keyword length, but the keyword length is not limited to 16 characters. The number of search results can be adjusted by changing the keyword length.

  ADVANTAGE OF THE INVENTION According to this invention, it has the effect that the optimal example which can be referred to when the person in charge of user support answers a question from a user can be presented. The reason is that one-byte characters included in the question sentence are extracted and used as keywords to search for cases without using knowledge and rules for keyword extraction and search result selection.

  The best mode for carrying out the present invention will be described with reference to the drawings.

  Referring to FIG. 1, a related case reference system 1 according to an embodiment of the present invention includes a terminal 10 that receives a question mail from a user and returns an answer to the question, and a trouble error related to the information processing system by the user. A question mail describing a message and an error log is sent to the terminal 10 in charge, and a user terminal 30 that receives answers, a case storage server 40 that includes a database 42 that stores searchable mail and answer mail cases in a searchable manner, And a network 20 connected to the network. The responsible terminal 10, the user terminal 30, and the case storage server 40 operate by program control.

  Referring to FIG. 2, the user terminal 30 includes a processing device 31 that connects an input device 32 such as a keyboard and an output device 33 such as a display. The processing device 31 includes data input / output means 311 for the input device 32 and output device 33, an input screen for inputting mail or mail processing means 312 for outputting mail, and mail transmission / reception means for transmitting / receiving mail to / from the network 20. 313.

  Referring to FIG. 3, the responsible terminal 10 includes a processing device 11 that connects an input device 12 such as a keyboard and an output device 13 such as a display. The processing device 11 receives the data input / output means 111 for the input device 12 and the output device 13, and the question mail from the network 20, displays it on the output device, and when the question mail is not assigned a question management number, The question management table 114 for registering the question management number attached to the question mail, the question mail processing means 112 for sending the question mail to the case storage server 40, and the search results sent from the case storage server 40 are arranged in descending order of the number of searches. A reply mail processing means 113 for displaying and outputting a predetermined number of cases from the top to the output device 13 and outputting an input screen for inputting a reply to the output device 13; .

  Referring to FIG. 4, the case accumulation server 40 extracts a keyword from a question mail, manages the keyword, the question management number of the extraction source, and the mail file name of the question mail, and handles the case of the search result based on the keyword. 10 and a database 42 for storing and managing cases so as to be searchable.

  The processing device 41 includes a mail transmission / reception unit 411 that transmits / receives a mail via a network, a keyword extraction unit 413 that generates a word string including a one-byte character string from the question mail and extracts a keyword, and a question management number of the question mail A query processing means 412 that makes it possible to search the database 42 based on the extracted keyword and the mail file name, a case search means 414 that obtains a case having a keyword extracted from the question mail from the database 42 and sends it to the terminal 10 in charge; And database management means 415 for managing the data for 42 so as to be searchable.

  The database 42 manages the mail management table 430 that manages the question mail and its reply mail by associating the question management number added to the question mail with the extracted keyword and the question management number of the question mail from which it is extracted. And a case database 470 that accumulates mail files of question mails and their answer mails.

  Next, the operation of the embodiment of the present invention will be described with reference to the drawings.

  Referring to FIG. 5, the user terminal 30 creates a question mail with the data input on the question mail input screen, and the question to the answer mail is attached with the question management number of the corresponding answer mail. A mail is created (step 51), and a question mail is transmitted to the responsible terminal 10 (step 52). The responsible terminal 10 receives and displays the question mail (step 53). When there is no question management number in the question mail, the question mail is a question mail transmitted for the first time from the user terminal 30, so the question management number is added and the question management number is registered in the question management table 114 ( Step 54) and further transfer to the case storage server 40 (Step 55).

  The case storage server 40 receives the question mail (step 56). The question management number attached to the question mail and the file name of the question mail are registered in the mail management table 430. If the question management number has already been registered in the mail management table 430, the mail file name is additionally registered in the corresponding question management number (step 57). The case accumulation server 40 extracts a keyword from the question mail (step 58) (the operation of keyword extraction will be described in detail in FIG. 7), attaches the question management number of the source question mail to the extracted keyword, Register in the search table 450. If the keyword is already registered in the search table 450, a question management number is additionally registered to the corresponding keyword (step 59). The case accumulation server 40 obtains a question management number corresponding to all the extracted keywords from the search table 450, obtains a mail file name corresponding to the question management number from the mail management table 430, and uses the corresponding mail file name as a case. The database 470 is searched (step 60), and the search result is transferred to the responsible terminal 10 (step 61). The responsible terminal 10 receives the search result (step 62).

  Next, referring to FIG. 6, the case accumulation server 40 obtains the question management number from the search table 450 for all keywords extracted from the question mail, and the case of the mail file name having the question management number is stored in the case database. Obtained from 470 and transmitted to the responsible terminal 10 (step 61), and the responsible terminal 10 receives the case that is the search result (step 62). The responsible terminal 10 selects a predetermined number from the received cases having the larger number of searches (step 63), and displays the selected case (step 64). When the person in charge refers to the displayed case and inputs the answer data, the terminal 10 in charge creates an answer mail based on the input answer data, and assigns a question management number corresponding to the question mail to the answer mail. (Step 66), it transmits to the user terminal 30. The user terminal 30 receives and displays the answer mail (step 68). Furthermore, the responsible terminal 10 transmits the answer mail to the case accumulation server 40 (step 69). The case storage server 40 receives the answer mail (step 70), additionally registers the mail file name of the answer mail in the corresponding question management number of the mail management table 430, and stores it in the case database 470 (step 71).

  Details of the keyword extraction operation in the case storage server 40 will be described with reference to FIG. The keyword extraction unit 413 of the processing device 41 obtains a 1-byte character string by excluding the 2-byte character and the line feed code from the head of the question mail body passed from the question mail processing unit 412 (step 581). Continuous 1-byte character strings are recognized as words, and the words are separated by spaces (step 582). A series of word strings ending with words in which the number of characters in the word string consisting of the one-byte character string word and the space between the words reaches a predetermined number or more is extracted as a keyword (step 583). The keywords starting with the word are sequentially extracted from the beginning of the word string, and the process is repeated until the keyword including the final word is extracted (step 584).

  With the above operation, a word is recognized from an error message or error log described in a 1-byte character string described in the question mail, and ends with a word that reaches a predetermined number of characters starting with a 1-byte character string word. A plurality of keywords are extracted, and the word strings of the keywords partially overlap with the word strings of other keywords.

  With reference to FIG. 8, the operation (step 59) of registering the extracted keyword in the search table 450 by the question mail processing means 412 will be described. The question mail processing means 412 registers all the keywords passed from the keyword extraction means 413 with the question management number of the question mail from which the keywords are extracted in the search table 450 (step 591). If the extracted keyword is already registered in the search table 450, a question management number is additionally registered for the corresponding keyword (step 592).

  With reference to FIG. 9, the operation of the case search means 414 searching for the case mail file stored in the case database 470 by the keyword extracted by the keyword extraction means 413 will be described. The case search unit 414 refers to the search table 450 for each extracted keyword, and obtains a question management number corresponding to the keyword from the search table 450 (step 601). The case search means 414 refers to the mail management table 430 by the question management number, acquires the mail file name to which the question management number is attached (step 602), searches the case database, and acquires related cases. (Step 603). By this operation, the question mail is searched for the number of times corresponding to the number of extracted keywords, and the case having a keyword matching the extracted keyword is searched for the number of times corresponding to the degree of matching. All cases of this number are sent to the terminal 10 in charge.

  With reference to FIG. 10, the operation | movement which selects the example received by the charge terminal 10 is demonstrated. The reply mail processing means 113 of the responsible terminal 10 selects cases (step 64). In the search results, the cases corresponding to the number of searches, that is, mails with a large number of searches are counted (step 641). A number of cases are selected and displayed (step 642). With the above operation, the keyword extraction source mail has the highest hit count, and mails having the same keyword in common are searched as similar cases.

  Next, the operation will be described using the exemplified data. The keyword extraction will be described using data illustrated in FIG. Keywords are extracted from the question mail 80 shown in FIG. In FIG. 11B, the question mail 80 is moved to a table 81 having row codes and column numbers for the convenience of explaining the extraction operation. The following operation will be described with reference to steps 581 to 584 in FIG. However, the keyword length is set to 16 characters or more.

  Step 581: The 2-byte character and the line feed code are removed from the head of the question mail 80. In the example of FIG. 11B, kanji and hiragana, and the line feed code at the end of the character string from line A to line F are to be removed. However, in the description, there is no problem even if it is not removed, so that the illustration of FIG.

  Step 582: The continuous 1-byte character string in FIG. 11B is an alphabetic character and a number, “(”, “)” parentheses, “:”, “.”, Recognize the continuous part as a word, separate the word with a space, and recognize the word string.

  Step 583: The number of characters is checked from the first word of the word string. If there are more than 16 characters, they are used alone as keywords, and a word string ending with a word that reaches more than 16 characters including spaces is extracted as a keyword.

  Step 584: Step 583 is repeated until a keyword ending with the last word in the word string is extracted.

  An operation for registering the keyword extracted by the above operation in the search table 450 will be described. Referring to FIG. 12, keywords 4511 to 4526 are registered in keyword 451 in the order of extraction. The question management number of the source question email corresponding to the keyword is registered in the search table 450 in association with the corresponding keyword. Each time a question mail is received, the extracted key wide 451 and the question management number 455 of the question mail from which it is extracted are registered in the search table 450. When the extracted keyword 451 is already registered in the search table 450, the corresponding question management number is additionally registered in the question management number 455 of the corresponding keyword.

  Further, the keyword extraction process will be described with reference to FIG. 11B and FIG.

  Since (A1 to A7), (A17 to A19), (A26), (A27), and (F2 to F30) in the table 81 of FIG. 11B are 2-byte characters, they are excluded.

  (A8 to A10), (A16), (B1), (B11), (B17), (B25), (C1), (C8), (C16), (C19), (D1), ( Since D8), (D12), (D15), (E1 to E3), (E140), (E24), and (F1) are spaces, they are excluded leaving one space.

  (A11 to A15), (A20 to A25), (B2 to B10), (B12 to B16), (B18 to B24), (B26 to B30), (C2 to C7), (C9 to C15) in the table 81 , (C17), (C18), (C20 to C25), (D2 to D7), (D9 to D11), (D13), (D14), (D6 to D21), (E4 to E9), (E11E23) , (E25 to E31) are recognized as words because they are continuous 1-byte characters.

  Since (A28), (B31), (C26), (D22), (E32), and (F31) in the table 81 are line feed codes, they are excluded.

  Next, a keyword in which the number of characters in the word string that separates the words with a space sequentially from the first word of the word string in the table 81 exceeds 16 characters is created.

  In the first keyword extraction, since the words “A11 to A15” in the table 81 are 5 characters, when combined with the words “A20 to A25” (6 characters) with a space between them, 12 characters are obtained, but still less than 16 characters. Therefore, it is stored in the search table 450 as a keyword 4511 of 22 characters including two spaces between the words “B2 to B10” (9 characters) and 20 characters.

  In the second keyword extraction, since the words “A20 to A25” in the table 81 are 6 characters, the word “B2 to B10” (9 characters) is inserted and a space is inserted between them to search as a keyword 4512 of 16 characters. Stored in the table 450 for use.

  In the third keyword extraction, since the words “B2 to B10” in the table 81 are 9 characters, the words “B12 to B16” (5 characters) are interspersed with 15 spaces, Since it is less than 16 characters, a space is inserted between the words “B18 to B24” (7 characters) and the total is stored in the search table 450 as a keyword 4513 of 24 characters.

  As described above, for all the words in the table 81, the keywords are combined in the order of the word string so that the words in the character string are 16 characters or more, and the keyword 4526 is created. The reason why the keyword length in this embodiment is 16 characters or more is that when the search is performed with a keyword exceeding 16 characters, the search result tends to decrease rapidly as illustrated in FIG. This is because the number of unnecessary search results is suppressed using this feature.

  Referring to FIG. 13, in the mail management table 430 for registering the mail file name of the mail having the same question management number as the question management number attached to the question mail, when the question mail is the first time, the question management number is the question management number. The same question management number is given to the answer mail, indicating that it is an answer mail for a specific question mail. The mail file name of the question mail itself is registered in the file name corresponding to the question management number 431. Further, when a question mail is further sent to the answer, the same question management number as that of the answer mail is given to indicate that it is a related mail, and the file corresponding to the question management number 431 in the mail management table 430 It is additionally registered in the name 435.

  Referring to FIG. 14, it is shown that question mail and answer mail files are stored in the case database 470.

  Thus, the operation of the embodiment of the present invention is completed.

  In the above-described embodiment, the question from the user is transmitted from the user terminal as a question mail, and the answer from the person in charge is transmitted from the terminal in charge as an answer mail. May be received by the terminal in charge, and converted into document data by a voice recognition technology that is an existing technology, and the above-described processing may be performed. In that case, instead of mailing an answer, the person in charge may be referred to and selected until the search result example is displayed on the terminal in charge.

  The first effect in the embodiment of the present invention is that all persons in charge can search and refer to the same case. The reason is that all the persons in charge do not need any special instruction or operation to select, and the same keyword is automatically extracted from the question mail itself, and the case can be searched and referred to.

  Furthermore, the second effect in the embodiment of the present invention is that the answers of all the persons in charge are made uniform and the user service is improved. The reason is that, because the case is automatically searched and referenced with the same keyword, the reference cases obtained from the search results are the same, and answers based on the same case are created.

  Furthermore, the third effect of the embodiment of the present invention is that the man-hours for knowledge and rule maintenance relating to the extraction of keywords related to the database of past cases and searches are not required. This is because keywords are extracted from question mails by routine operations without using knowledge or rules for keyword extraction processing.

  Furthermore, the fourth effect in the embodiment of the present invention is that case accumulation can be automated. The reason is that the question mail and the answer mail for the question mail can be managed by the question management number and can be made into a case database.

  The present invention can be applied to a reception business system of a support center for an information processing system.

It is a figure which shows the structure of embodiment of this invention. It is a figure which shows the structure of the user terminal 30 in embodiment of this invention. It is a figure which shows the structure of the charge terminal 10 in embodiment of this invention. It is a figure which shows the structure of the case storage server 40 in embodiment of this invention. It is a figure which shows the operation | movement which sends the question mail from the user terminal 30 in embodiment of this invention to the case storage server 40. FIG. It is a figure which shows the operation | movement which produces the reply mail of the charge terminal 10 in embodiment of this invention with reference to the case of the case storage server 40. FIG. It is a figure which shows the keyword extraction operation | movement by the keyword extraction means 413 in embodiment of this invention. It is a figure which shows the keyword registration operation | movement by the question mail process means 412 in embodiment of this invention. It is a figure which shows the case search operation | movement by the case search means 414 in embodiment of this invention. It is a figure which shows case selection operation | movement of the search result by the reply mail process means 113 in embodiment of this invention. A diagram (a) is a diagram illustrating a question mail, and a diagram (b) is a diagram for explaining a process of moving a question mail to a table and performing keyword extraction. It is a figure which illustrates the table 450 for a search which matches a keyword, a question management number, and a mail management number. It is a figure which illustrates the mail management table 430 which matches a question management number and a mail file. It is a figure which illustrates the example database 470 which accumulate | stores the mail file relevant to a question mail. It is a figure which illustrates the relationship between keyword length and the number of search results.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Related case reference system 10 Responsible terminal 12 Input device 13 Output device 20 Network 30 User terminal 40 Case accumulation server 41 Processing device 42 Database 111 Input / output means 112 Question mail processing means 113 Answer mail processing means 114 Question management table 115 Mail transmission / reception means 411 Mail transmission / reception means 412 Question mail processing means 413 Keyword extraction means 414 Case search means 415 Database management means 430 Mail management table 450 Search table 470 Case database

Claims (10)

Keyword extracting means for inputting question data, extracting a one-byte character string from the question data, and using it as a keyword;
Case search means for searching and reading case information corresponding to the keyword from a case database;
A case reference system characterized by comprising Output the case information read by the case search means, and input the answer data corresponding to the case information;
Database management means for associating the answer data input by the answer processing means with the question data and storing it as the case information in the case database;
The case reference system according to claim 1, further comprising: The keyword extracting unit extracts a word string having a predetermined number of characters or more from the question data as a keyword by extracting a word string of a plurality of one-byte character strings including spaces between words. 2. The case reference system described in 2. A keyword extraction step in which the case accumulation server inputs the question data, extracts a one-byte character string from the question data, and uses it as a keyword;
A case search step in which the case storage server searches and reads case information corresponding to the keyword from a case database;
A case reference method characterized by including: A terminal device that outputs the case information read in the case search step and inputs the answer data corresponding to the case information; and
A database management step in which the case accumulation server associates the answer data input in the answer processing step with the question data and stores it in the case database as the case information;
The case reference method according to claim 4, further comprising: The keyword extraction step extracts a word string having a predetermined number of characters or more from the question data as a keyword by extracting a word string of a plurality of one-byte character strings including spaces between words. 5. The case reference method according to 5. Receives the question data transmitted from the terminal device, extracts a 1-byte character string from the question data as a keyword, reads out the case information corresponding to the keyword from the case database, and transmits it to the terminal device And a database management means for receiving answer data corresponding to the case information from the terminal device and storing the answer data in the case database as the case information in association with the question data. Case accumulation server characterized by that. 8. The keyword extracting unit extracts a word string having a predetermined number of characters or more from the question data as a keyword by extracting a word string of a predetermined number of characters from a plurality of one-byte character strings including spaces between words. Case storage server. Receives the question data transmitted from the terminal device, extracts a 1-byte character string from the question data and uses it as a keyword, reads out case information corresponding to the keyword from the case database, and sends it to the terminal device And a database management process for receiving answer data corresponding to the case information from the terminal device and storing the answer data in the case database as the case information in association with the question data. A program characterized by having it be performed. 10. The keyword extraction process according to claim 9, wherein a word string having a predetermined number of characters or more is extracted from the question data for a plurality of one-byte character string word strings including spaces between words, and used as the keyword. Program.

Images