JP2001117942A - INFORMATION SEARCHING DEVICE, INFORMATION SEARCHING METHOD, AND COMPUTER-READABLE RECORDING MEDIUM RECORDING PROGRAM FOR CAUSING COMPUTER TO EXECUTE THE METHOD - Google Patents

Abstract

(57) [Summary] [Problem] In an information search for a plurality of information sources, a clue for selecting an appropriate information source for an operator by finely calculating the relevance of each information source to a search request. To provide. A search file storage unit that stores a search file created for each database, a reception unit that receives a search request from a client, a request processing unit that determines the type of search request, A document relevance calculation unit 303 that calculates the relevance of each document to the search request based on the similarity between the vectors, and a document extraction unit 304 that extracts a predetermined number of documents based on the calculated relevance.
A database suitability calculation unit 305 for calculating the suitability of each database based on the number of extracted documents for each database and the sum of the suitability, and a transmission unit 306 for transmitting the processing result to the client. .

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieval apparatus, an information retrieval method, and a computer-readable recording recording a program for causing a computer to execute the information retrieval method from among a plurality of databases. Regarding the medium.

[0002]

2. Description of the Related Art Computerized documents and data are stored in a storage device such as a hard disk as a database, and when a search request is input from an operator, information matching the search request is searched. 2. Description of the Related Art An information search device that outputs information on a display screen or the like has been conventionally known.

[0003] In some cases, the above-mentioned device can execute a search by designating any one of several different databases provided inside the device or inside another information processing device or the like connected via a network. Many. In that case, the operator must specify which database among the available databases is to be searched.

[0004] By the way, as the number of available databases increases, the probability of obtaining target information increases, but on the other hand, which database should be designated as a search target, in other words, the target information is stored in which database. Presence of existence becomes difficult.

[0005] It is good for the operator to know in advance the outline of the information stored in each database based on experience, a usage manual, or the like, but for example, a database in which daily newspaper articles are stored mechanically. In some cases, almost no one can tell the whole picture of the information stored inside.

[0006] In such a database, it is impossible to judge whether target information is likely to be obtained or not unless a database is first specified and searched. However, as the newspaper article databases described above are typical, such databases are sometimes huge and can take a long time to search.

In order to prevent such trial and error at the stage of specifying a search object, there is known an information search apparatus for automatically selecting an appropriate database from input search words and executing a search. I have.

For example, according to the invention described in Japanese Patent Application No. 4-122988, a list of keywords included in documents in the available database is prepared in advance, and the entered search keywords are stored in the list. By comparison, if the keyword is present in the list, a search based on the keyword is executed for the database that is the base of the list.

[0009]

However, in the above-mentioned prior art, if a search keyword appears at least once in a database, the database is used as a search target by the keyword.

Therefore, if an actual search is performed, 10
There is no distinction between database A, which can obtain zero matching documents, and database B, which can obtain only one document. If you only want to see some dozens of samples as a sample, for that purpose, only database A is searched, but database B is also searched. There is a problem that a complicated process occurs.

Further, for example, when a search is performed using “wine” and “French cuisine” as keywords for the purpose of investigating southern French cuisine using wine, a database A that stores documents relating to French cuisine as a whole is also obtained. As a whole, the database B, which stores documents on French current affairs, is also searchable as long as it has such keywords. Therefore, the search results include literature on the wholesale price of wine and literature on the shortage of human resources in the French cuisine industry. There is a problem that noise is mixed in and a desired document is hardly found.

Further, according to the above-mentioned prior art, it is possible to avoid wasting time and labor due to a so-called missed search (the number of hits becomes zero). However, although the number of compatible documents is not zero, the database is extremely small. In some cases, it may not be possible to improve the processing efficiency by selecting a database.

[0013] For example, the larger the database, the higher the possibility of including a search keyword in the database. Therefore, the above-mentioned newspaper article database and the like are always searched for unless the keyword is a special technical term. It is considered that the possibility is high.

Therefore, when many of the available databases contain many terms that cover various fields, such as newspaper articles, most of the databases are eventually searched according to the above-mentioned prior art, and as a result, processing is difficult. There are problems such as a long time, a high cost (when a fee is charged for searching the database), and an enormous amount of search results.

[0015] The above-mentioned problem is to determine whether or not a certain database is a target of the search, in other words, whether or not a certain database matches the purpose of the search, by simply determining whether or not there is a search keyword. The determination is made based on two values of nonconformity, and the operator is not given a chance to correct the determination result.

According to the present invention, in order to solve the above-described problem of the related art, in information search for a plurality of information sources, the degree of suitability of each information source with respect to a search request is finely calculated, so that an operator can be appropriately searched. It is an object of the present invention to provide an information search device, an information search method, and a computer-readable recording medium storing a program for causing a computer to execute the method, which can provide a clue for selecting a suitable information source.

[0017]

Means for Solving the Problems To solve the above-mentioned problems,
In order to achieve the object, an information retrieval apparatus according to the present invention is an information retrieval apparatus for selecting a database that matches a retrieval purpose from among a plurality of databases, wherein an input of a document stored in the plurality of databases is performed. First calculating means for calculating the relevance to the searched search request for each document; and a predetermined number of documents stored in the plurality of databases based on the relevance calculated by the first calculating means. Extracting means for extracting a document; second calculating means for calculating, for each database, the relevance of the plurality of databases to the search request based on the number of documents extracted by the extracting means for each database;
It is characterized by having.

According to the first aspect of the present invention, the relevance of each database to the search request is finely determined based on the number of documents of each database in the documents extracted based on the relevance to the search request. Is calculated.

According to a second aspect of the present invention, there is provided an information search apparatus for selecting a database that matches a search purpose from among a plurality of databases, wherein an input of a document stored in the plurality of databases is performed. First calculating means for calculating the relevance to the search request for each document, and a predetermined number of documents from the documents stored in the plurality of databases based on the relevance calculated by the first calculating means. Extracting means for extracting, based on the relevance calculated by the first calculating means of the document extracted by the extracting means, calculating the relevance to the search request of the plurality of databases for each database 2 calculation means.

According to the second aspect of the present invention, the relevance of each database to the search request is determined based on the sum or average of the relevance of each database of the documents extracted based on the relevance to the search request. Calculated finely based on

According to a third aspect of the present invention, in the information search method for selecting a database that matches a search purpose from among a plurality of databases, an input of a document stored in the plurality of databases is performed. A first calculation step of calculating the relevance to the search request for each document, and a predetermined number of documents from the documents stored in the plurality of databases based on the relevance calculated in the first calculation step. And a second calculation step of calculating, for each database, the relevance of the plurality of databases to the search request based on the number of documents extracted by the extraction step for each database,
It is characterized by including.

According to the third aspect of the present invention, the relevance of each database to the search request is finely determined based on the number of documents in each database in the documents extracted based on the relevance to the search request. Is calculated.

According to a fourth aspect of the present invention, there is provided an information search method for selecting a database matching a search purpose from among a plurality of databases, wherein an input of a document stored in the plurality of databases is performed. A first calculation step of calculating the relevance to the search request for each document, and a predetermined number of documents from the documents stored in the plurality of databases based on the relevance calculated in the first calculation step. Extracting, based on the relevance calculated in the first calculation step of the document extracted in the extraction step, calculating the relevance of the plurality of databases to the search request for each database 2 calculation steps.

According to the fourth aspect of the present invention, the relevance of each database to the search request is determined based on the sum or average of the relevance of each document in the document extracted based on the relevance to the search request. Calculated finely based on

According to a fifth aspect of the present invention, there is provided a recording medium storing a program for causing a computer to execute the method according to the third or fourth aspect.
The program becomes machine-readable, whereby the operation of claim 3 or 4 can be realized by a computer.

[0026]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of an information retrieval apparatus, an information retrieval method and a computer-readable recording medium according to the present invention are described below with reference to the accompanying drawings. The form will be described in detail.

(Embodiment) First, a system configuration of an information search system including an information search device according to an embodiment of the present invention will be described. FIG. 1 is a block diagram showing a system configuration of an information search system including an information search device according to an embodiment of the present invention.

In FIG. 1, reference numeral 101 denotes an information retrieval apparatus according to an embodiment of the present invention. The information search device according to the embodiment of the present invention also doubles as an information search device. That is, as will be described later, it has a function of not only presenting a database that matches the search purpose of the operator, but also performing a normal information search on the database and presenting a document that meets the search request.

Reference numerals 102, 103 and 104 denote file servers which hold a document database and transmit a designated document to a request source in response to a request. "Political article database" with accumulated information, "International article database" with documents on current events outside Japan
have.

The documents stored in these databases are stored in a predetermined file format, specifically, an S-file.
It is unified into JIS format. What is S-JIS format SG
It has a format similar to a structured document such as ML, for example, <b
ody> tag, the document body, <title> tag, document title, <author> tag, document creator name, <createddate> tag, document creation date, <pu
Following the bdate> tag, the date and time of publication of the document (date and time of publication in the publication, etc.) are described.

When a document of a different format, for example, a document of a unique format created by commercially available word processing software is added to these databases, the file format is converted prior to the addition. Information retrieval device 10 described above
Reference numeral 1 has a function (filter) for converting files of various formats into the S-JIS format, but a detailed description thereof is omitted since it is not the main content of the present invention.

The file servers 102, 103, and 104 hold a plurality of search files used by the information search device 101 to search for those documents, in addition to the document files. These search files describe information such as the frequency of occurrence of all words included in the documents in the database and the documents in which they appear.

Reference numerals 105 and 106 are used to input various requests from the operator via a keyboard or the like, and
This is a client that outputs the processing result input from the server to a display or the like. The information retrieval apparatus 101, file servers 102, 103 and 104, client 105
And 106 are connected by a network 100, respectively.

In the embodiment of the present invention, the information retrieval device 1
01 and the file servers 102, 103 and 104 are provided separately, but the information retrieval apparatus 101 itself may manage the database and have a function as a file server. The information search device 101 may have functions as the clients 105 and 106 so that the information search device 101 can input various requests and output results.

The information retrieval apparatus 101 is connected to the file servers 102, 103, 104 and the clients 105,
106 may also be used. In this case, the input of the search request and the output of the processing result are performed in the information search apparatus 101, and the processing such as the calculation of the degree of matching described later is performed by the database or the document held by the information search apparatus 101 itself. Will be performed.

The information retrieval device 101, the file servers 102, 103 and 104, and the clients 105 and 106 are specifically a workstation,
It is realized by a personal computer or the like.

Next, a hardware configuration of the information retrieval apparatus 101 according to the embodiment of the present invention will be described.
FIG. 2 is a block diagram illustrating a hardware configuration of the information search device according to the embodiment of the present invention.

2, reference numeral 201 denotes a CPU for controlling the entire system; 202, a ROM storing a boot program and the like; 203, a RAM used as a work area of the CPU 201; A hard disk drive (HDD) 205 controls reading / writing of data from / to a hard disk 205, and an HD 205 stores data written under the control of the HDD 204.

Reference numeral 206 denotes an FDD (floppy disk drive) for controlling reading / writing of data from / to an FD (floppy disk) 207 under the control of the CPU 201; Are respectively shown.

Reference numeral 208 denotes a display for displaying windows related to data such as documents, images, and function information, including a cursor, icons, and tool boxes.
An interface (I / F) 209 is connected to the network NET via the communication line 210 and controls an interface between the network NET and the inside.

A keyboard 211 has a plurality of keys for inputting characters, numerical values, various instructions, etc.
Is a mouse for moving a cursor and selecting a range, or moving and resizing a window, selecting and moving an icon, 213 is a scanner for optically reading an image, and 214 is a printer for printing contents displayed in a window. 215 is a CD-ROM which is a removable recording medium.
A ROM 216 indicates a CD-ROM drive for controlling reading of data from the CD-ROM 215. Reference numeral 200 denotes a bus for connecting the above components.

Next, a functional configuration of the information retrieval apparatus according to the embodiment of the present invention will be described. FIG. 3 is a block diagram functionally showing the configuration of the information search device according to the embodiment of the present invention.

Referring to FIG. 3, the information retrieval apparatus according to the embodiment of the present invention includes a retrieval file storage unit 300, a reception unit 301, a request processing unit 302, a document relevance calculation unit 303, and a document extraction unit. The configuration includes a 304, a database compatibility calculation unit 305, and a transmission unit 306. Also, the document relevance calculating unit 303 includes the query vector creating unit 3
03a, a document vector creation unit 303b, and a vector similarity calculation unit 303c.

The search file storage unit 300 acquires a search file of a database to be searched from a file server that manages the database and stores it. Then, the search file once stored is not deleted unless the storage capacity of the search file storage unit 300 is insufficient, so that the next time the same database is searched, if the search file remains here, There is no need to obtain the file from the file server again, and the processing efficiency is improved.

Here, for convenience of explanation, all the search files of the “economic article database”, “political article database”, and “international article database” existing in the file servers 102, 103 and 104 are stored in the search file storage section 300. Is stored.

The receiving unit 301 receives various requests from the client 105 or 106 via the network 100 according to a predetermined protocol. The received request is output to a request processing unit 302 described later.

The request processing unit 302 determines the type of the request input from the receiving unit 301. As described above, since the information search device according to the embodiment of the present invention also serves as an information search device, there are roughly the following two types of requests to be input.

(1) Presentation of a database suitable for the purpose of search This request includes at least a search sentence (a search word or a search expression, such as “About bank consolidation and merger” In the form, an inquiry is made using a normal natural sentence). Processing for this request is the main content of the present invention, and details will be described later.

(2) Presentation of Document Suitable for Search Purpose This request includes at least a search sentence, the name of a database to be searched, and other additional information such as a period to be searched. This is a normal information search request, and instructs to output the text of the conforming document in the designated database, or its title, abstract, and the like.

In addition to the above, there are conceivable requests for maintenance of a search environment (including setting of a user password, addition / deletion of an available database, addition / deletion of a document to / from a database), and the like. Since the content is not typical, detailed description is omitted.

The type of the request input from the receiving unit 301 is
The determination is made by referring to the value of a specific bit indicating the type in the request. Alternatively, at least the above (1) or (2) may be determined based on the presence or absence of a database name. That is, if a database is specified together with a search sentence (2), if there is a search sentence but no database is specified (1).

Then, in accordance with the type of request determined in this way, necessary information is output to an appropriate function unit. Specifically, when the input request is the above (1) or (2), a search sentence or the like included in the request is transmitted to a document relevance calculating unit 303 described later, and the type of the request is transmitted to a document described later. Output to the extraction unit 304.

The document suitability calculating section 303 corresponds to “first calculating means” in the claims, and includes a query vector creating section 303a, a document vector creating section 303b,
A vector similarity calculation unit 303c.

The query vector creation unit 303a creates a query vector that numerically expresses its meaning based on the search sentence input from the request processing unit 302.
The structure of the vector will be described later.

The document vector creation unit 303b creates a document vector of each document stored in the database based on the search file of each database stored in the search file storage unit 300.

Here, the document vector is a vector having the same number of element values as all the phrases contained in all the documents in the database, and the semantic content of the document is determined by the characteristic of each element value corresponding to each phrase. Can be grasped numerically.

Simply, each element value of the document vector is determined by the frequency at which the corresponding phrase appears in the document. For example, if the phrase in the document appears once, the element value in the document vector corresponding to the phrase becomes 1, if it appears 10 times, it becomes 10, and if it does not appear at all, it becomes 0.

However, for example, even if a document contains many words "bank", if another document in the same database also contains many words, the degree to which the word is characteristic for the document is determined. It is considered low. Conversely, even if a document contains only a few words "bank", if the word does not appear in any other document, then the word may simply characterize the document when compared to other documents. It is thought that it is expressed in.

In view of such circumstances, in the embodiment of the present invention, each element value of a vector is determined not by the frequency of appearance of a simple phrase, but by the statistical characteristics of the appearance of the phrase, that is, between documents, Alternatively, the calculation is performed in consideration of the distribution status in one document. As described above, such information is described in the search file created for each database. Therefore, the document vector creation unit 303b refers to these search files cached in the search file storage unit 300. To create a document vector for each document.

The vector similarity calculator 303c sequentially calculates the similarity between the query vector created by the query vector creator 303a and the individual document vectors created by the document vector creator 303b. The vector similarity is specifically calculated by a predetermined formula based on the inner product thereof.

Since each vector is a numerical representation of the meaning of a search sentence or a document, the similarity of the vector is used to determine the search sentence that is the source of the query vector and the document that is the source of the document vector. The similarity of the semantic content, in other words, the relevance of the document to the search sentence.

When the relevance of each document to the search text is calculated as described above, the document relevance calculating unit 303 associates each document with the database to which it belongs, and the relevance for the search text, and The list (or a location where the list exists, for example, an address on a memory) sorted in the descending order of
04 is output. FIG. 4 shows the document relevance calculating unit 30.
FIG. 4 is an explanatory diagram schematically showing an example of a list created by No. 3.

The document extracting unit 304 determines a predetermined number of input documents based on the type of request input from the request processing unit 302 and the list of relevance of each document input from the document relevance calculating unit 303. To extract documents. The extraction criterion and the output destination of the extracted document differ depending on the type of request.

That is, when the type of the request input from the request processing unit 302 is the above (1), the document extracting unit 304 refers to the list input from the document relevance calculating unit 303 and Some documents, for example, the top five documents are extracted in descending order. In the example of FIG. 4 described above, five documents are extracted from “alliance between Bank A, Bank B, and Bank C” to “rumors of alliance between Bank D and Bank G in the United States”. Then, extract the record of the extracted document from the above list,
Output to the database adaptability calculating unit 305 described later.

When the type of request is the above (2), for example, ten documents having a high relevance are shown, and in the example of FIG. 4, "Affiliation between Bank A, Bank B and Bank C" is changed to "US D
Extract the documents up to "Bargaining between Bank and E Bank". Then, the record of the extracted document is extracted from the list and output to the transmission unit 306 described later.

The database relevance calculating unit 305 corresponds to the “second calculating unit” in the claims, and calculates the relevance of each database from the document extracted by the document extracting unit 304 to the input search text. Is calculated. This calculation is specifically performed by focusing on the following numerical values.

(A) Number of documents for each database For example, in the example of FIG. 4, among the top five documents extracted by the document extraction unit 304, the documents of the "economic article database" are "bank A, bank B, Bank C tie-up, "Bank A, B
Two documents, "Ripple of Bank and C Bank Alliance", and "Political Article Database" document, "Promoting Consolidation of Government and Bank"
There are two documents in the "International Article Database": "The merger problem between Bank D, Bank E and Bank F in the United States" and "Rumors about the partnership between Bank D and Bank G in the United States".

If this number is regarded as the relevance of each database, the databases with the higher relevance are sequentially “economic article database” and “international article database” (same rank),
Next is the "political article database". According to this calculation method, a database having many documents with high relevance has a higher rank.

(B) Sum or average of the degrees of relevance of documents for each database For example, in the example of FIG. 4, the documents of the "economic article database" in the top five documents "Alliance of Bank A, Bank B and Bank C" The total relevance of "Ripple of Bank A, Bank B and Bank C alliance" is 243.579, the average value is 121.789 (rounded down), and the document "Political article, The sum and average of the fitness values for "Promote Consolidation" is 107.678, and the document "International Article Database" on "A merger problem between Bank D, Bank E and Bank F of the United States"
The total relevance of the "rumors of US D-bank and G-bank alliance" is 20
8.134, average value is 104.067.

Assuming that the sum of the degrees of matching of documents is the degree of matching of the database, the order of the database is “economic article database”, “international article database”, and “political article database” in descending order. It is an economic article database, a political article database, and an international article database. According to the calculation method using summation,
Even if the relevance of each document is relatively low, the ranking of the database with the large number of cases rises, and according to the calculation method using the average value, the database of the database with the relatively high relevance of each document even if the number of cases is small is high. The ranking will rise.

(C) Order of Documents in Each Database For example, the sum of values obtained by dividing the constant 1000 by the order of each document is defined as the relevance of each database. In the example of FIG.
"Economic article database" documents "Bank A, Bank B, C
Bank Alliance "" Ripple of Bank A, Bank B and Bank C Alliance "
Are 1st and 2nd respectively, so 1000/1 + 10
00/2 = 1500 is set as the fitness of the database.

Similarly, the relevance of the “political article database” is 1000/4 = 250, and the relevance of the “international article database” is 1000/3 + 1000/5 = 533.3 (rounded down). Article database, "international article database," and "political article database." According to this calculation method, a database having a document with a high degree of relevance even if the number of cases is small has a higher degree of relevance of itself.

Although the degree of conformity of the database may be calculated using any of the above numerical values, there is a case where one calculation method does not give any advantage (see the example of (a) above).
You may make it calculate combining some. Here, for the sake of convenience of explanation, the conformity of the database is “243.579” for the “economic article database” and “economic article database” based on the sum of the relevance of each document in (b) above.
Is 107.678, and "international article database" is 20
It is assumed that 8.134 is calculated.

After calculating the fitness of each database as described above, the database fitness calculating unit 305 outputs the names of the databases to the transmitting unit 306, which will be described later, in descending order of their rank. The value of the matching degree may be output together with the name of the database.

The transmitting unit 306 converts the processing result input from the document extracting unit 304 or the database matching degree calculating unit 305 into the network 100 according to a predetermined protocol.
To the client 105 or 106 via the.
The transmitted processing result is transmitted to the client 105 or 10
6 is output on the display screen.

FIG. 5 shows the client 105 or 106
Is an example of the processing result finally displayed on the display screen of the client 105 or 106 when the request input from the above is the request of the above-described (1) including the search sentence "about bank consolidation and merger". FIG.

Note that the degree of compatibility may be displayed together with the name of the database. Further, a search sentence input area and the like are further displayed on the screen of FIG. 5, and the operator inputs a new search sentence in the area, designates one of the databases with a mouse or the like, and clicks an execution button or the like. By clicking, a search with a new search sentence for the specified database may be executed.

When the above execution button or the like is clicked,
The client 105 or 106 creates the above request (2) based on the search sentence extracted from this screen and the name of the database, and transmits the request to the information search device 101. With this configuration, since a document can be searched for the database searched directly from the database search screen, operations such as screen switching are not required, and operability is improved.

In the embodiment of the present invention, the above-mentioned (1)
Is input, only the names of the databases are displayed in the order of higher relevance. Therefore, when the operator performs a subsequent search for a document, any one of the databases is designated and the above-mentioned (2) is renewed. However, it is also possible to automatically specify some top databases and then perform a document search using the same search sentence.

The search file storage unit 300, the receiving unit 301, the request processing unit 302, the document relevance calculating unit 303,
The document extracting unit 304, the database matching degree calculating unit 305, and the transmitting unit 306 are provided in the ROM 202 and the RAM 2
03 or the hard disk 205, the floppy disk 207, or the like, the CPU 201 or the like executes the instruction processing in accordance with the instructions described in the program recorded on a recording medium such as the recording medium, thereby realizing the function of each unit.

Next, a series of processes of the information search device (also as an information search device) according to the embodiment of the present invention will be described. FIG. 6 is a flowchart illustrating a processing procedure of the information search device according to the embodiment of the present invention.

In the flowchart shown in FIG. 6, first, in step S601, the client 301
Receive the request sent from 05 or 106. In subsequent step S602, the request processing unit 302 determines whether the request input from the receiving unit 301 is a presentation request of a database suitable for the search purpose.

If the input request is a request to present a database (Yes at step S602), at step S603, the query vector generating unit 303a generates a query vector from the input search query. In the following step S604, the document vector 303
At b, create a document vector for each document in the database.

In step S605, the vector similarity calculator 303c calculates the similarity between the query vector and each document vector, and outputs a list of the calculation result to the document extractor 304 in step S606.

In step S607, the document extracting unit 304 extracts a predetermined number of documents (this number is predetermined according to the type of request) from the input list in the order of relevance. Then, in step S608, the database matching degree calculation unit 305 determines in step S6
The relevance of each database is calculated based on the document extracted in 07.

Further, in step S609, transmitting section 306
, The names of the databases are transmitted to the client 105 or 106 in descending order of the degree of matching calculated in step S608, and the processing according to this flowchart ends.

If the input request is not a database presentation request in step S602 (No in step S602), in step S610, the request input from the receiving unit 301 is a document presentation request suitable for the search purpose. Is determined.

When the input request is a document presentation request (Yes at step S610), step S611 is performed.
If the request is not a document presentation request (No at Step S610), the process according to this flowchart ends, and another process according to the type of the request is performed.

Steps S611 to S615 correspond to step S
The description is omitted because it is the same as 603 to S607. In step S616, the transmitting unit 306 executes step S616.
The title of the document extracted in 615 and the degree of conformity thereof are transmitted to the client 105 or 106, and the processing according to this flowchart ends.

As described above, according to the embodiment of the present invention, the relevance of a certain database to a certain search request is finely calculated by a numerical value that takes a continuous value instead of a binary value of conformity or non-conformity. The user can refer to the result and select the database as needed, eliminating the need to search for unlikely databases or unnecessarily large numbers of databases. Is improved.

The operator first narrows down the database with a simple search sentence and then performs a search with a more complicated search sentence, so that a large number of documents stored in a large number of databases can be searched for search noise and system noise. It is possible to efficiently search for a desired document while keeping the processing load low.

[0092] Also, when the retrieval of documents is not the main purpose, but rather when one wants to roughly know the characteristics and trends of information stored in available information sources, the information retrieval apparatus according to the present invention can be used. Can be. For example, find out which newspaper companies are actively reporting on a particular case, and which of the various electronic bulletin boards connected by the network have more information on a particular matter. If you want. ADVANTAGE OF THE INVENTION According to this invention, the characteristic of each information source can be easily grasped only by inputting the inquiry by a natural sentence.

The information search method described in the embodiment of the present invention is realized by executing a prepared program on a computer such as a personal computer or a workstation. This program is for hard disk, floppy disk, CD-RO
The program is recorded on a computer-readable recording medium such as M, MO, and DVD, and is read out from the recording medium and executed by the computer. This program can be distributed via the recording medium or as a transmission medium via a network such as the Internet.

[0094]

As described above, according to the first aspect of the present invention, the first calculating means determines the relevance of the documents stored in the plurality of databases to the input search request. The extraction means extracts a predetermined number of documents from the documents stored in the plurality of databases based on the fitness calculated by the first calculation means, and the second calculation means Calculating the relevance of the plurality of databases to the search request for each database based on the number of documents extracted by the extraction means for each database, so that the relevance of each database to the search request is Calculated finely based on the number of documents in each database in the documents extracted based on the degree of relevance to An effect that is capable of information retrieval device to provide clues for selecting an information source.

According to the second aspect of the present invention, the first calculating means calculates, for each document, the relevance of the documents stored in the plurality of databases to the input search request and extracts the document. Means for extracting a predetermined number of documents from the documents stored in the plurality of databases based on the degree of matching calculated by the first calculating means, and second calculating means for extracting by the extracting means Said first of said document
Based on the relevance calculated by the calculating means, the relevance of the plurality of databases to the search request is calculated for each database, so that the relevance of each database to the search request is based on the relevance to the search request. Based on the sum and average of the relevance of each extracted database for each database, it can be calculated finely, providing the operator with clues for selecting an appropriate information source. This provides an effect that a simple information retrieval device can be obtained.

According to the third aspect of the present invention, the first calculating step calculates, for each document, the relevance of the documents stored in the plurality of databases with respect to the input search request, and extracts the document. A step of extracting a predetermined number of documents from the documents stored in the plurality of databases based on the degree of matching calculated by the first calculation step, and a second calculation step by the extraction step Based on the number of documents for each database, to calculate the relevance of the plurality of databases to the search request for each database, the relevance of each database to the search request,
It is finely calculated based on the number of documents in each database in the documents extracted based on the relevance to the search request, thereby providing the operator with clues for selecting an appropriate information source. There is an effect that an information search method capable of performing the search can be obtained.

According to the fourth aspect of the present invention, the first calculating step calculates the degree of conformity of the documents stored in the plurality of databases to the input search request for each document, and extracts the document. A step of extracting a predetermined number of documents from the documents stored in the database based on the degree of matching calculated in the first calculation step, and a second calculation step extracted by the extraction step Based on the relevance calculated in the first calculation step of the document, the relevance of the plurality of databases to the search request is calculated for each database. Of documents extracted based on the degree of relevance to each database, is calculated finely based on the sum or average value of relevance for each database. Information retrieval method capable of providing a clue to select appropriate information sources is an effect that can be obtained.

According to the fifth aspect of the present invention, by recording a program for causing a computer to execute the method according to the third or fourth aspect, the program becomes machine-readable. Accordingly, an effect is obtained that a recording medium capable of realizing the operation of claim 3 or 4 by a computer can be obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a system configuration of an information search system including an information search device according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a hardware configuration of the information search device according to the embodiment of the present invention.

FIG. 3 is a block diagram functionally showing a configuration of the information search device according to the embodiment of the present invention.

FIG. 4 is an explanatory diagram schematically showing an example of a list created by the document relevance calculating section 303 according to the embodiment of the present invention.

FIG. 5 is an explanatory diagram showing an example of a display on the client 105 or 106 of a processing result of the information search device according to the embodiment of the present invention.

FIG. 6 is a flowchart illustrating a processing procedure of the information search device according to the embodiment of the present invention;

[Explanation of symbols]

 100 Network 101 Information Retrieval Device 102, 103, 104 File Server 105, 106 Client 200 Bus 201 CPU 202 ROM 203 RAM 204 HDD 205 HD 206 FDD 207 FD 208 Display 209 I / F 210 Communication Line 211 Keyboard 212 Mouse 213 Scanner 214 Printer 215 CD-ROM 216 CD-ROM drive 300 Search file storage unit 301 Receiving unit 302 Request processing unit 303 Document relevance calculation unit 303a Query vector generation unit 303b Document vector generation unit 303c Vector similarity calculation unit 304 Document extraction unit 305 Database Fitness calculation unit 306 Transmission unit

Claims (5)

[Claims] 1. An information search apparatus for selecting a database that matches a search purpose from among a plurality of databases, wherein a relevance of a document stored in the plurality of databases to an input search request is calculated for each document. First calculating means, extracting means for extracting a predetermined number of documents from the documents stored in the plurality of databases based on the degree of matching calculated by the first calculating means, and extracting by the extracting means And a second calculating means for calculating, for each database, the relevance of the plurality of databases to the search request based on the number of documents obtained for each database. 2. An information search apparatus for selecting a database that matches a search purpose from among a plurality of databases, wherein the relevance of a document stored in the plurality of databases to an input search request is calculated for each document. First calculating means, extracting means for extracting a predetermined number of documents from the documents stored in the plurality of databases based on the degree of matching calculated by the first calculating means, and extracting by the extracting means Second calculating means for calculating, for each database, the relevance of the plurality of databases with respect to the search request based on the relevance calculated by the first calculating means for the obtained document. Information retrieval device. 3. An information search method for selecting a database that matches a search purpose from a plurality of databases, wherein a degree of relevance of a document stored in the plurality of databases to an input search request is calculated for each document. A first calculation step; an extraction step of extracting a predetermined number of documents from the documents stored in the plurality of databases based on the fitness calculated in the first calculation step; A second calculation step of calculating, for each database, the relevance of the plurality of databases to the search request based on the number of documents obtained for each database. 4. An information search method for selecting a database that matches a search purpose from among a plurality of databases, wherein a degree of relevance of a document stored in the plurality of databases to an input search request is calculated for each document. A first calculation step; an extraction step of extracting a predetermined number of documents from the documents stored in the plurality of databases based on the fitness calculated in the first calculation step; And a second calculation step of calculating, for each database, the degrees of relevance of the plurality of databases to the search request based on the degrees of relevance calculated by the first calculation step of the obtained document. Information retrieval method. 5. A computer-readable recording medium on which a program for causing a computer to execute the method according to claim 3 or 4 is recorded.