JP4146067B2 - Document search system and document search method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a document search system such as a natural language search system and a document search method used in the search system.
[0002]
[Prior art]
Conventionally, a keyword search system is well known as a document search method. This keyword search system searches for a document including a keyword input by a user. In order to search for a target document using this system, the user needs to specify an appropriate keyword included in the target document. In this case, especially when there are a large number of documents in the database, a large number of candidate documents can be obtained simply by specifying a single keyword. Therefore, the user further refines the candidates by additionally inputting keywords. A refined search is required.
[0003]
As a system using such a search method, a full-text search system for searching for a document file required by a user from a large number of document files has been developed. In this document search system, in order to narrow down search candidates, the user needs to connect several words with “or”, “and”, or “not”, create a search expression, and search accordingly. .
[0004]
Therefore, recently, a natural language search system capable of efficiently searching registered document data with daily words (natural language) has been developed. In a natural language search system, just by inputting a question sentence (query) in everyday language (natural language), the query content analysis, syntax analysis, morpheme analysis, keyword weighting, and similarity calculation are performed automatically. Thus, an appropriate document can be easily retrieved from the database.
[0005]
Such a natural language search system requires advanced document analysis processing. Therefore, the search target document file is analyzed in advance to generate search data corresponding thereto, and the search data is registered in the database. It is necessary to keep it.
[0006]
[Problems to be solved by the invention]
However, even if a natural language search system is used, since the search is performed on a word basis, whether or not useful information is written in the searched document is determined by the user himself / herself. It was necessary to browse and read the contents. In particular, when a plurality of candidate documents are searched for a certain question sentence, it is necessary for the user to browse the candidate documents in order until it reaches useful information.
[0007]
In addition, on the search result list display screen, candidate documents are displayed in descending order of similarity (score order) with similarities to the question sentences, and the similarities are calculated on a word basis, and therefore the same Documents that contain many words tend to have high scores. For this reason, a document in which useful information for a user is written does not necessarily have a high score, and there are many cases in which useful information is written in a low-score document.
[0008]
The present invention has been made in view of the above-described circumstances, and can provide a user with an index for determining the degree of usefulness regarding a searched document file, and a document in which useful information is written can be obtained. It is an object to provide a document search system and a document search method that can be easily obtained by a user.
[0009]
[Means for Solving the Problems]
In order to solve the above-described problems, the present invention provides a document search system for searching a document file registered in a database, and a document file registered in the database in accordance with a keyword or a question sentence input from a user. A search means for searching the document file, the evaluation information input by the user for the document file searched by the search means, and the keyword or question sentence used for searching the document file And means for registering in the database, and information presentation means for presenting the evaluation information and the keyword or question sentence registered in association with the document file for each document file searched by the search means It is characterized by comprising.
[0010]
In this document retrieval system, the evaluation information input by the user for the retrieved document file is registered in the database in association with the document file together with the keyword or question text used to retrieve the document file. By doing so, for each search document displayed on the search result list screen or the document browsing screen, it is possible to present the corresponding evaluation information together with the keyword or question text that is the basis of the evaluation. it can. Therefore, as an index for determining the degree of usefulness of the retrieved document file, it is shown to the user what kind of evaluation is obtained when using a keyword or a question sentence. Therefore, the user can easily obtain a document in which useful information is written.
[0011]
As the input evaluation information, a document registered in the database using information selected from a plurality of types of evaluation determined in advance according to the difference in usefulness of the retrieved document file is used. It is preferable to further provide means for counting the number of inputs of each evaluation type for each file and presenting the input evaluation type and the count value of the input number to the user. As a result, it is possible to notify the user of the total result for each evaluation type, so that it is possible to more easily determine the degree of usefulness related to the retrieved document file.
[0012]
Further, by providing a means for ranking and presenting the usefulness of each document file searched by the search means based on the count result of the number of inputs of each evaluation type, it is more than using only a simple score display. It is possible to efficiently obtain a highly useful document.
[0013]
When a document file is searched by the search means, the search means searches the similarity between the keyword or question sentence used for the search and the keyword or question sentence already registered in the database. The search based on the means for calculating each document file, the calculated similarity, and the evaluation type of the evaluation information registered corresponding to each document file searched by the search means By providing means for ranking and displaying the usefulness of each document file searched by means, not only the evaluation information but also the similarity between the entered keyword or question sentence and the already registered keyword or question sentence The ranking can be performed in consideration of the degree, and the ranking of the usefulness can be more accurately performed.
[0014]
Further, the present invention provides a document retrieval system for retrieving a document file registered in a database, search means for retrieving a document file registered in the database according to a keyword or a question sentence input from a user, Means for inputting evaluation information for the document file searched by the search means, and means for ranking and presenting the usefulness of the document file registered in the database based on the input evaluation information It is characterized by comprising. Thus, based on the input evaluation information, it is possible to determine the degree of usefulness related to the retrieved document file only by the configuration in which the usefulness of the document file registered in the database is ordered and presented. The indicator can be presented to the user.
[0015]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 shows a functional configuration of a natural language search system according to an embodiment of the present invention. The natural language search system 111 is a search system that can efficiently search a pre-registered document file with daily words (natural language) and keywords, and questions from users with daily words (natural language). A sentence (query) is analyzed, and further, a syntactic analysis, a morphological analysis, a keyword weighting process, a similarity calculation, and the like are performed to search for an appropriate document from registered documents. The natural language search system 111 includes a natural language search database (DB) 201, a document registration unit 202, a search / list display unit 203, an original text extraction unit 204, an evaluation recording unit 205, an evaluation rank display unit 206, as shown in FIG. An evaluation similarity calculation unit 207 is provided.
[0016]
The natural language search database (DB) 201 stores document files registered as search targets and search data (index data) corresponding thereto. Further, in this embodiment, the natural language search database (DB) 201 is provided with an additional field for each document file, and the evaluation information input from the user for the document file is searched for the document file. It is configured so that it can be registered in association with the document file together with the keyword or question sentence used in the above. As the evaluation information, evaluation type information such as “useful”, “useless”, “other”, and a comment sentence that can be freely input by the user are used.
[0017]
The document registration unit 202 is for registering a document file to be searched in the natural language search database (DB) 201. For example, the document registration unit 202 periodically reads a document file from a predetermined file storage unit or the like, and stores the document file. Search data (index data) is generated and registered in the natural language search database (DB) 201. At the time of document registration, a predetermined initial value is set in the additional field. The registered document file is managed by a unique document ID.
[0018]
The search / list display unit 203 searches the document file registered in the natural language search database (DB) 201 in accordance with a search sentence (keyword or natural language question sentence) input from the user, and performs the search. A list of candidate documents obtained by the above is provided to the user as a search result list. In this case, on the search result list screen, together with a list of candidate documents, a total result for each evaluation type calculated by the evaluation similarity calculation unit 207 is displayed for each candidate document.
[0019]
The original text retrieval unit 204 retrieves the original text of the document file selected by the user on the search result list screen from the file storage unit or the natural language search database (DB) 201 and provides it to the user. In this case, the evaluation information and the search text (keyword or question text) corresponding to the selected document file are also taken out and provided to the user together with the above-described total result for each evaluation type.
[0020]
The evaluation recording unit 205 associates the evaluation information input by the user on the browse screen of the searched document file with the corresponding document file together with the search sentence (keyword or question sentence), and the natural language search database (DB). ) 201 is registered in the additional field 201. The evaluation rank display unit 206 ranks the retrieved documents based on the evaluation information for each retrieved document.
[0021]
In this natural language search system 111, the evaluation information for the document can be input from the original text browsing screen after the document search, and the evaluation information can be displayed together with the search text based on the evaluation. . Therefore, it is possible to present to the user what kind of evaluation is obtained when using which keyword or question sentence, and this is the degree of usefulness regarding the retrieved document file. It can be used as an index for judging Specifically, the following functions are provided.
[0022]
(1) An evaluation type / comment input item is provided on the retrieved document browsing screen so that the evaluation of the document can be input.
(2) A place for displaying the input evaluation is provided on the browsing screen of the retrieved document so that the evaluation on the document can be browsed.
(3) An additional field for storing the evaluation is provided, and the evaluation is stored in the database 201 together with the document.
(4) A mechanism for counting the number of inputs for each evaluation type is provided, and the evaluation is counted for each document.
(5) Display the total number of evaluations on the search result list screen and the browsing screen.
(6) At the time of storing the evaluation, an additional field for recording the question sentence or the keyword that was used when inputting the evaluation is provided for each type of evaluation, and the question sentence or the keyword is recorded there.
[0023]
(7) When displaying the evaluation corresponding to each document, the question text or keyword that is the basis of the evaluation is displayed together with the type of evaluation, and what kind of inquiry is evaluated is displayed.
(8) Using the input evaluation, the usefulness of registered documents is ranked and displayed.
(9) A new calculation is performed using the input evaluation and the recorded question sentence or keyword, and a similarity calculation reflecting the evaluation is performed.
[0024]
FIG. 2 shows the configuration of a client server type computer system to which the natural language search system 111 of FIG. 1 is applied. As shown in the figure, this computer system includes a server computer 11 and a plurality of client computers 12. Server computer 11 and client computer 12 are connected via a computer network such as a LAN or the Internet.
[0025]
As shown in the figure, the server computer 11 is provided with the above-described natural language search system 111 and an HTTP server 112. The natural language search system 111 includes a search server 301 and a search engine 302 in addition to the natural language search database (DB) 201 described above. A search engine 302 is a software module that performs the processing for the natural language search described above, and realizes the function of the search / list display unit 203 in FIG. The search server 301 is a software module for providing a search service to the client computer 12 through the HTTP server 112. Here, the document registration unit 202, the original text extraction unit 204, the evaluation recording unit 205, and the evaluation order display in FIG. Functions such as a unit 206 and an evaluation similarity calculation unit 207 are provided.
[0026]
The user of each client computer 12 can access the server computer 11 through the Web browser 121 to search and browse the document file, input and browse the evaluation, and the like.
[0027]
(Database field structure)
Next, the field (column) structure of the natural language search database (DB) 201 will be described with reference to FIG.
In the natural language search database (DB) 201, a record is assigned to each registered document, and information necessary for search, input evaluation information, and the like are stored in columns in the record. In document search, it can be set for each column whether or not to search.
[0028]
In columns 1 to n, index data for search related to a certain registered document is stored. In subsequent columns n + 1 to n + m, count values of the number of inputs for each evaluation type (1 to m) are stored. For example, when three types of “useful”, “useless”, and “others” are used as evaluation types, as shown in FIG. 4, m = 3, and these are input for each of the three types of evaluations. The number of times will be held in the corresponding column.
[0029]
In the subsequent columns n + m + 1 to n + m + m, comments input for each evaluation type (1 to m) are registered. A plurality of comments can be registered in each column. Further, in the subsequent columns n + 2m + 1 to n + 2m + m, a question sentence (or keyword) based on the evaluation is registered for each evaluation type (1 to m). A plurality of question sentences (or keywords) can be registered in each column.
[0030]
(Document registration process)
Next, the procedure of document registration processing executed by the document registration unit 202 will be described with reference to the flowchart of FIG.
First, a document file to be registered is read from a file storage unit or the like (step S101). Then, after initial setting is performed for the additional fields (columns n + 1 to n + 2m + m) of the registration destination record (step S102), the search data (index) obtained by the morphological analysis for the document file is the columns 1 to n. (Step S103). In the initial setting process for the additional fields (columns n + 1 to n + 2m + m), as shown in FIG. 6, “0” is set as the count value of the number of input times in the columns n + 1 to n + m, and the comment registration columns n + m + 1 to n + m + m and question sentence registration columns n + 2m + 1 to n + 2m + m are set to be empty.
[0031]
(Evaluation registration process)
Next, evaluation information and question sentence registration processing will be described with reference to the flowchart of FIG.
[0032]
When a user inputs a question sentence (or keyword) on a Web page provided to the Web browser 121 by the search server 301 and a search operation is requested by a user operation (step S11), the input question sentence The search / list display unit 203 executes processing for searching candidate documents from the database 201 based on (or keywords), and a search result list screen including a list of candidate documents is provided to the Web browser 121 that is the search request source ( Step S111).
[0033]
When a user selects a document to be viewed on the search result list screen, a document acquisition request including the document ID is issued (step S12), and the document acquisition request is sent to the original text extraction unit 204. The original text retrieving unit 204 retrieves the document file specified by the document ID and provides it to the requesting Web browser 121 as a registered document browsing screen (step S112). On the registered document browsing screen, as shown in FIG. 8, the registered document designated by the document ID is displayed. In addition, on this registered document browsing screen, as input locations for evaluation information, radio buttons for inputting evaluation types (“useful”, “useless”, “other”), comment input fields, and registration of evaluation information are registered. There is a “Register” button to specify. The user can read the registered document and select “useful”, “useless”, or “other” by using the radio button for inputting the evaluation type. Then, after the user inputs a comment text in the comment input field as necessary, the user inputs an evaluation by pressing a “register” button (step S13).
[0034]
The input evaluation type and comment are sent to the evaluation recording unit 205, and are registered in the additional field of the record in which search data for the corresponding document ID is registered together with the question sentence or keyword used in the search execution process. (Step S113).
[0035]
(Evaluation registration process)
Next, the procedure of the evaluation registration process executed by the evaluation recording unit 205 will be described with reference to the flowchart of FIG.
[0036]
First, the document ID related to the document to be evaluated and the evaluation information input from the user are read (step S121). FIG. 10 is an example of evaluation information input from the user. Here, it is assumed that information “useful” is input as an evaluation type and a comment sentence “useful information is written” is input. Next, the information already registered in the additional field corresponding to the document ID is read (step S122), and the contents of the evaluation information read in step S121 are added to the information to update the additional field (step S122). S123, S124). In the case of FIG. 10, the count value of the input count of the field corresponding to the “useful” evaluation type is incremented by 1, and “useful information is written” in the field corresponding to the “useful” evaluation type. Will be additionally registered. In addition, the question sentence “How to write a guide letter?” Used for the search is additionally registered.
[0037]
(Presentation of evaluation information)
The registered evaluation information is displayed on the registered document browsing screen as shown in FIG. As shown in FIG. 11, a display area for presenting an evaluation list is provided on the registered document browsing screen, and as shown in FIG. : 2, “useless”: 0, “others”: 0) and an evaluation type, a comment, and a question sentence are displayed for each registered evaluation information. Further, the user name that has entered the evaluation information is recorded in the database 201 as the registrant name, so that the registrant name may be displayed together.
[0038]
(Update additional fields)
Next, the additional field update operation will be described in detail with reference to FIGS. 13 and 14.
[0039]
FIG. 13 shows the contents of the additional field before adding the evaluation information, and FIG. 14 shows the contents of the additional field after additionally registering the evaluation information of FIG.
The value of the field corresponding to the column n + 1, that is, the “useful” evaluation type is updated from the previous count value 3 to 4, and the field in the column n + m + 1 corresponding to the “useful” evaluation type is registered so far. Comment sentences are recorded by separating them with “,”, and a comment sentence “Useful information is written” is newly registered there. Also, in the column n + 2m + 1 corresponding to the “useful” evaluation type, the question texts registered so far are separated by “,”, and there is a question “How to write a guide letter?” A new sentence is newly registered.
[0040]
(Similarity display # 1)
Next, processing for ranking and presenting the usefulness of each candidate document displayed on the search result list screen based on the registered evaluation information will be described. In this example, the normal similarity (score) indicated by the number of words that match the search sentence (question sentence or keyword) is recalculated based on the evaluation information, and is used as ranking information for usefulness. Assume a case. In this case, the search / list display unit 203 executes the process shown in the flowchart of FIG.
[0041]
That is, first, a search sentence (question sentence or keyword) input by the user is read (step S131), and a process for searching candidate documents from the database 201 based on the search sentence is performed (step S132). In this search processing, the contents of columns 1 to n of each record in the database 201 are used, and the ID of the candidate document (hit document) including the word requested in the search sentence and the score of the document are acquired (step S133). ). Thereafter, for each hit document, the count value (evaluation count) of the number of inputs for each evaluation type is extracted from the corresponding columns n + 1 to n + m (step S134), and the score is recalculated based on the evaluation count. Then, according to the recalculation result, each candidate document is ranked and displayed on the search result list display screen (step S135). That is, in the score recalculation process, as shown in FIG. 17, the evaluation count is extracted for each evaluation type (step S141), and the evaluation score is calculated by multiplying it by the coefficient determined for each evaluation type ( In step S142), a process of adding it to the score and displaying it (step S143) is performed. An example of the search result list display screen in this case is shown in FIG.
[0042]
As shown in FIG. 16, in the search result list display screen, for each rank of usefulness, the similarity (recalculated score), the title of the corresponding candidate document, and the evaluation count for the candidate document are displayed. Is displayed.
[0043]
(Similarity display # 2)
Next, not only the evaluation type but also the similarity between the keyword or question text used in the search and the keyword or question text recorded for each evaluation type is also displayed on the search result list screen. A process for ranking and presenting the usefulness of each candidate document will be described. In this case, the search / list display unit 203 executes the process shown in the flowchart of FIG.
[0044]
That is, first, a search sentence (question sentence or keyword) input from the user is read (step S151), and a process for searching candidate documents from the database 201 based on the search sentence is performed (step S152). In this search processing, the contents of columns 1 to n of each record in the database 201 are used, and the ID of the candidate document (hit document) including the word requested in the search sentence and the score of the document ID (ID = i) ( (Similarity = Si search) is acquired (step S153). Thereafter, for each hit document, the similarity (Si) between the input search sentence and the already registered search sentence. ^j ) Is repeatedly executed while changing the evaluation type j (steps S154 and S155), whereby a new score (similarity) is recalculated (step S156).
[0045]
In the recalculation of the score considering the similarity of the search sentences, the formula shown in FIG. 19 is used. That is, the similarity Si for the search sentence for a document with a document ID = i is a normal score (Si search) calculated from the number of words including the word specified in the search sentence and the similarity based on the evaluation (Si Is given by the function). In this case, the similarity based on the evaluation (Si evaluation) is the similarity (Si) of the search sentence for each evaluation type j. ^j ) Is the coefficient (C ^j ) Is multiplied and expressed as the sum of them. Coefficient (C ^j For example, as shown in FIG. 20, weighting values such as +0.2 are used for the evaluation type “useful” and −0.1 is used for the evaluation type “useless”.
[0046]
Through the above processing, ranking can be performed considering not only evaluation information but also the similarity between the entered keyword or question sentence and the already registered keyword or question sentence. Can be ranked.
[0047]
For example, when an evaluation of “useful (j = 1)” is input to a document with ID = 20, the question sentence is recorded in the column n + 2m + 1. Thereafter, when the document with ID = 20 is searched. , C ^j Since is positive, the score including the evaluation becomes high. On the other hand, if an evaluation of “useless (j = 2)” is entered, ^j Since the is negative, the score with the evaluation is low. When the degree of similarity of the question sentences is low, the value of the evaluation (Si evaluation) itself added to the score is low, and it becomes possible to rank the usefulness levels more accurately reflecting the evaluations on the question sentences. .
[0048]
In the similarity display # 1 and # 2, the ranking of usefulness is performed by recalculating the normal score, but only the evaluation or the similarity and the evaluation of the question sentence without using the score. It is also possible to rank the usefulness by using only.
[0049]
In the above description, the case where the document search system of the present embodiment is applied to a client-server type computer system has been described as an example. However, the document search system can be similarly applied to a stand-alone computer. .
[0050]
Although the case where the evaluation information and the question sentence are presented on the document browsing screen has been mainly described, the evaluation information and the question sentence can be displayed for each document on the search result list display screen. In addition, since the natural language search system 111 according to the present embodiment is realized by a computer program, the computer program is stored in a computer-readable storage medium so that the computer program can be transferred to a normal computer through the storage medium. The effect similar to that of the present embodiment can be obtained only by introduction. The computer program can be distributed not only via a storage medium but also via a communication medium.
[0051]
Further, the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the scope of the invention in the implementation stage. Further, the above embodiments include inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some constituent requirements are deleted from all the constituent requirements shown in the embodiment, the problem described in the column of the problem to be solved by the invention can be solved, and the effect described in the column of the effect of the invention Can be obtained as an invention.
[0052]
【The invention's effect】
As described above, according to the present invention, it is possible to present a user with an index for determining the degree of usefulness related to a searched document file, and a document in which useful information is written can be presented to the user. Can be easily obtained.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a functional configuration of a document search system according to an embodiment of the present invention.
FIG. 2 is an exemplary block diagram illustrating a configuration of a client server type computer system to which the document search system according to the embodiment is applied.
FIG. 3 is an exemplary view showing a field structure of a database used in the document search system according to the embodiment;
FIG. 4 is an exemplary view showing a relationship between database column positions and evaluation types used in the document search system of the embodiment;
FIG. 5 is an exemplary flowchart illustrating the procedure of document registration processing in the document search system according to the embodiment;
FIG. 6 is a view showing an example of initial values of evaluation information registered in the document search system of the embodiment.
FIG. 7 is an exemplary flowchart illustrating a procedure of registration processing of evaluation information and a question sentence in the document search system according to the embodiment;
FIG. 8 is a view showing an example of a registered document browsing screen used in the document search system of the embodiment.
FIG. 9 is an exemplary flowchart illustrating the procedure of evaluation registration processing in the document search system according to the embodiment;
FIG. 10 is a diagram showing an example of evaluation information input from a user in the document search system of the embodiment.
FIG. 11 is an exemplary view showing an example of a registered document browsing screen used in the document search system of the embodiment.
12 is a diagram showing an example of the contents of an evaluation list display unit displayed on the registered document browsing screen of FIG.
FIG. 13 is a view showing an example of the contents of an additional field before evaluation information is added in the document search system of the embodiment.
FIG. 14 is a view showing an example of the contents of an additional field after evaluation information is added in the document search system of the embodiment.
FIG. 15 is an exemplary flowchart illustrating a procedure of a usefulness ranking process which is executed in the document search system according to the embodiment;
FIG. 16 is an exemplary view showing an example of a search result list display screen used in the document search system of the embodiment.
FIG. 17 is an exemplary flowchart illustrating a procedure of score recalculation processing in the document search system according to the embodiment;
FIG. 18 is a flowchart showing another example of a usefulness ranking process in the document search system according to the embodiment;
FIG. 19 is a view for explaining an operation for score recalculation in the document search system according to the embodiment;
FIG. 20 is a view showing an example of weight values used in the document search system of the embodiment.
[Explanation of symbols]
111 ... Natural language search system
201 ... Natural language search database
202 ... Document registration section
203 ... Search / list display section
204 ... Original text retrieval unit
205 ... Evaluation recording section
206 ... Evaluation rank display section
207 ... Evaluation similarity calculation unit
11 ... Server computer
12 ... Client computer
111 ... Natural language search system
112 ... HTTP server

Claims (3)

In a document search system that searches document files registered in the database,
Search means for searching a document file registered in the database according to a keyword or a question sentence input from a user;
Evaluation information composed of evaluation types selected from a plurality of predetermined evaluation types input from the user for the document file searched by the search means, and for searching the document file Means for registering the used keyword or question sentence in the database in association with the document file;
Information presentation means for presenting the evaluation information and the keyword or question sentence registered in association with the document file for each document file searched by the search means ;
When a document file is searched by the search means, the search means searches the similarity between the keyword or question sentence used in the search and the keyword or question sentence already registered in the database. Means for calculating each document file,
Based on the calculated similarity and the evaluation type of the evaluation information registered corresponding to each document file searched by the search means, the usefulness of each document file searched by the search means is calculated. A document search system comprising ranking display means for ranking and displaying . In a document search method for executing a search for a document file registered in a database by a document search system,
A search step in which the document search system searches for a document file registered in the database according to a keyword or a question sentence input by a user;
Evaluation information composed of an evaluation type selected from a plurality of predetermined evaluation types, which is input from the user to the document file searched by the search step by the document search system, Registering the keyword or question sentence used for searching the document file in the database in association with the document file;
An information presentation step in which the document search system presents the evaluation information and the keyword or question sentence registered in association with the document file for each document file searched in the search step;
When the document search system searches for a document file in the search step, the similarity between the keyword or question sentence used for the search and the keyword or question sentence already registered in the database, Calculating for each document file searched by the search step;
The document search system is searched by the search step based on the calculated similarity and the evaluation type of the evaluation information registered corresponding to each document file searched by the search step A document search method comprising: a ranking display step of ranking and displaying the usefulness of each document file. A computer-readable storage medium storing a computer program for searching a document file registered in a database, the computer program comprising:
A procedure for searching a document file registered in the database according to a keyword or a question sentence inputted by a user;
Evaluation information composed of an evaluation type selected from a plurality of predetermined evaluation types input from the user for the searched document file, and used for searching the document file A procedure for registering keywords or question sentences in the database in association with the document file;
An information presentation procedure for presenting the evaluation information and the keyword or question sentence registered in association with the document file for each retrieved document file;
When the document file is searched, the similarity between the keyword or question sentence used in the search and the keyword or question sentence already registered in the database is calculated as A procedure for calculating each retrieved document file;
Based on the calculated similarity and the evaluation type of the evaluation information registered corresponding to each retrieved document file, the usefulness of each retrieved document file is ranked and displayed. A storage medium that causes a computer to execute a ranking display procedure.

Images