JP2751793B2 - Information retrieval system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a natural language interface for information retrieval means such as a database.
The information requester can search in natural language without being aware of the existence of multiple databases, and by adjusting the database selection rules, the search of the database accumulated by information search specialists The present invention relates to an information search system that can use know-how efficiently.

[0002]

2. Description of the Related Art With the development of database technology, various large-scale commercial databases and in-company databases have been constructed. Since the information requester who wants to use this information does not necessarily know the respective search method and target area, in order to perform an actual search, it is often the case that a database search expert (searcher) performs a search. Was. However, the number of searchers has not kept up with the spread of databases.

Further, with the development of computer technology, information requesters who are not familiar with computers often use computers.

[0004] As a result of these two points, there has been an increasing demand for an interface that allows even information requesters who are not familiar with database search to easily search.

[0005] As one of the interfaces to respond to this demand, an interface for inquiring a computer of a database in a natural language has been developed. Such a natural language interface has a semantic analysis unit that performs natural language processing, understands the meaning of the input sentence of the input natural language, and follows the operation means unique to the database for each database. Create input columns and search the database.

[0006] In the conventional natural language interface, there is no framework for processing the characteristics of each database, the superiority between the databases, the inclusion relation, and the like. Therefore, the database is limited to a single database or a plurality of databases. Only systems that perform searches without distinction have been proposed. However, when a conventional method is used to search a plurality of databases having overlapping target areas, a redundant search is often performed or lack of completeness, which often does not meet the search intention of the information requester. There were many.

[0007] This is because the relationship between the target area and the database as performed by the searcher and the difference in the search knowledge for each database are not incorporated.

[0008]

A database that grasps the intention of an information requester by analyzing and processing a search request and dynamically changes the priority of database selection based on the relationship between a target area and the database. Using the selection rules, the search database selection unit accessing the database feature database selects a database and, for each database, selects a parameter for generating a search formula, so that each of the plurality of databases can be selected. In this way, the optimum search intended by the information requester can be executed.

[0009]

In order to solve the above-mentioned problems, an information search system according to the present invention is an information search system having a plurality of databases, the search request for analyzing a search request input from an information requester. The analysis unit, the target area knowledge describing knowledge for selecting a database, and the search request analysis unit by referring to the target area knowledge, the result of the analysis by the search request analysis unit into a search request format that framed the elements of the search A search request processing unit to be converted, a database feature database describing the comprehensiveness of the plurality of databases with respect to a target area, a database selection rule for selecting a database that meets a search request from the plurality of databases, Referring to the target area knowledge and the database feature database, Accordingly, a search database selection unit that selects a database that matches the search request format, and a search formula that matches the search request format is created and targeted for the database selected by the search database selection unit. It has a search formula generation unit that searches a database, receives and outputs search results from the database, and a search result presentation unit that presents the search results.

Further, the information search system of the present invention has a database search generation rule for searching the respective databases and the search expression generator for the plurality of databases, and selects the plurality of databases by the search database selector. A search formula generation unit starting unit that outputs the search request format to the search formula generation unit corresponding to the selected database, sets the database search generation rule, and starts the search formula generation unit; and A search result evaluation unit for removing duplicate data from the plurality of output search results.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings.

FIG. 1 is a diagram showing a basic configuration of an embodiment of the present invention, FIG. 2 is a diagram showing an example of a conceptual structure analyzed by a search request analysis unit in FIG. 1, and FIG. 3 is a search request processing unit in FIG. FIG. 4 is a view showing an example of a search request format processed by the client 103; FIG.
FIG. 4 is a diagram showing an example of a database feature database 104 in FIG. 1, and FIG. 5 is a diagram showing an example of a database selection rule 105 in FIG.

In FIG. 1, a search request analysis unit 101 analyzes a search request from an information requester and converts it into a conceptual structure 201. To parse the search request and convert it to a conceptual structure,
For example, IEICE Technical Report NLC-91-62
"Natural Language Interface Construction Kit: IF-Kit"
May be used.

The target area knowledge 102 has coverage of each database for the target area in the form of a network. For example, in the case of a document database, in the target area knowledge 102, the concept of the target area is organized by a synonymous relation having the same meaning and a hierarchical relation indicating the narrowness of the semantic range. It describes the knowledge of which concept a particular magazine contains.

The search request processing unit 103 receives the result of analysis from the search request analysis unit 101 and
The search request is converted into a search request format 3 while referring to FIG.
Convert to 01.

The search request format 301 includes a search information source type, a search subject, a reference item, a search range, a target language,
It consists of a search request level. The search information source type specifies media such as journal articles, patents, and books.

The search subject is a theme which is a key when searching. The reference item is an item on the database indicating information that the information requester actually searches and obtains, such as an abstract or a bibliographic item. The search range is, for example, “1
975 to the present ". The target language indicates what language is obtained by the search and described. The search request level indicates the exhaustiveness of the search, such as whether it is desired that all search requests are satisfied or whether only two or three representative requests are required.

The search request processing unit 103 includes a search request analysis unit 1
01 is converted into a search request format 301, and a part which is not clearly expressed in the request of the information requester and is not filled up is appropriately questioned or analogized to the information requester to obtain the search request format 301. Is determined, and a search request format 302 in which each parameter is filled is created.

The database feature database 104 includes:
It is a database of the features of each database, and consists of items such as subject area, recording information source type, recording time, number of records, recording language, and recording items.

The database selection rule 105 is a rule for selecting an appropriate database for a search request from a plurality of databases.

The search database selection unit 106 receives the search request format from the search request processing unit 103 and receives the target area knowledge 102 and the database feature database 10.
4 and the database selection rule 106, the database to be actually searched is selected.

The search formula generation unit starting unit 107 starts a search formula generation unit for each database selected by the search database selection unit 106, and uses a search formula generation rule corresponding to each database. Set to.

The search expression generators 108, 110, 112
A search formula is generated for the selected database, a search is executed, a search result from the database is received, and the search result is sent to the search result evaluation unit 114.

The search result evaluation unit 114 determines whether to wait for the search results from the plurality of search expression generation units 109 and 111 according to the search request format 302 sent from the search expression generation unit activation unit, and performs the search. The search result is sent to the search result presentation unit 113 by performing a process of removing duplicate data from the result.

The search result presentation unit 115 presents the search result sent from the search result evaluation unit 114 to the information requester.

Here, an example will be described. For the sake of explanation, it is assumed that a search request has been sent from an information requester to “show all journal articles related to information search.”

The search request analysis unit 101 analyzes the search request to create a conceptual structure 201, and the search request processing unit 103
Pass to The search request processing unit 103 creates a search request format 301 from the conceptual structure 201 with reference to the target area knowledge 401 in FIG. From FIG. 2, since “information search” in the analysis result is a concept in the target area knowledge 401,
The “information search” is set as a search subject. Further, the search information source type can be "magazine article" and the search request level can be "no leakage". Thus, the conceptual structure 201
The search request format created from is shown in FIG. Further, the search request processing unit asks the information requester about the reference item, search range, and target language of the part not buried in the search request format, or substitutes a default value to obtain the information of the search request format. The search request format 302 is created by filling in the unfilled parameters, and the search database selection unit 10
Pass to 6.

The search database selecting unit 106 receives the search request format 302 and selects a database to be searched according to the database selection rule 601 in FIG. 6 with reference to the target area knowledge 401 in FIG. 4 and the database feature database 501 in FIG. I do.

First, there are four databases containing the search information source type "magazine articles": database c, database d, database e, and database f. All of these databases have "informatics" as the subject area. Further, looking at the target area knowledge 2 in FIG. 2, "information search" exists as a subordinate concept of "informatics". Can be searched.
In addition, the four databases satisfy the reference items “bibliographic items + abstract”, target language “Japanese”, and search range “from 1975 to the present”.

Further, in the case of a journal article satisfying the search request level "no omission", the database contains journals dealing with the search subject and the concept related thereto.

Referring to the target area knowledge 401 shown in FIG. 4, the magazines dealing with the information search “information search” are A, B, and E.
The three. The four magazines A, B, C, and E deal with the concepts "information management", "document search", and "DB search" that have a hierarchical relationship with "information search". From the database feature database 501 shown in FIG. 5, the database A handles database A, the database f handles the magazine A, the database c and the database e handles the magazine B, the database d handles the magazine C, and the magazine E. Can be determined as the database f.

Here, the search database selection unit 106 selects a database so as to cover four magazines A, B, C, and E with as few databases as possible.

Here, a database containing two or more magazines is a database c and a database f. These two databases are selected because the magazines contained therein do not completely overlap. The database d includes only the magazine C, but the database d is also selected because there is no other database that includes the magazine C. The database e contains only the magazine B, but if the database c is selected for the magazine B, there is no need to select the database e.

Therefore, the search database selection unit 10
6 is database c, database d, database f
Select

List of databases selected by the search database 6 (database c, database d, database f)
From the search request format 302, the search formula generation unit starting unit 107 generates the search formula generation unit 108 having the search generation rule 109 for the database c, the search formula generation unit 110 having the search generation rule 111 for the database d, and the database f
And a search expression generation unit 112 having a search generation rule 113 for use, and changes the search request format to a search result evaluation unit 114.
Send to

The search formula generation unit 108 generates a search formula from the request format 302 so as to conform to the format of the search formula of the database c (116), sends the search formula to the database c (116), and returns from the database c. The search result and the search result evaluation unit 114 are sent.

The search formula generation unit 110 generates a search formula from the request format 302 according to the format of the search formula of the database d (117), sends the search formula to the database d (117), and returns from the database d. The search result and the search result evaluation unit 114 are sent.

The search formula generation unit 112 generates a search formula that conforms to the format of the search formula of the database f (118) from the request format 301, sends the search formula to the database f (118), and returns from the database f (118). The search result is sent to the search result evaluation unit 114.

The search result evaluation unit 114 matches the respective search results sent from the search expression generation units 108, 110 and 112 according to the search request format sent from the search expression generation unit activation unit, and Except for what seems to be possible, it is sent to the search result presentation unit 115.

The search result presenting section 115 presents the search result sent from the search result evaluating section 114 to the search requester.

Although the present invention has been specifically described based on the embodiments, the present invention is not limited to the above-described embodiments, and it is needless to say that various changes can be made without departing from the gist of the present invention. Nor.

[0042]

As described above, according to the present invention, the information requester can perform his / her requested search without being aware that a plurality of databases exist. By adjusting the database selection rules, database search know-how accumulated by information search specialists can be used efficiently.

[Brief description of the drawings]

FIG. 1 is a basic configuration diagram showing one embodiment of the present invention.

FIG. 2 is a diagram illustrating an example of an analysis result of a search request unit.

FIG. 3 is a diagram illustrating an example of a search request format.

FIG. 4 is a diagram showing an example of search target area knowledge.

FIG. 5 is a diagram showing an example of a database feature database.

FIG. 6 is a diagram illustrating an example of a database selection rule.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 101 Search request part analysis means 102 Target area knowledge 103 Search request processing part 104 Database feature database 105 Database selection rule 106 Search database selection part 107 Search expression generation part starting part 108 Search expression creation knowledge for database c 109 Search expression generation for database c Part 110 Knowledge for creating search formulas for database d 111 Search formula creation unit for database d 112 Knowledge for creating search formulas for database f 113 Search formula creation unit for database f 114 Search result evaluation unit 115 Search result presentation unit 116 Database c 117 Database d 118 Database f 201 Example of conceptual structure 301 Example of search request format that is not completely filled 302 Example of search request format that is completely filled 401 Example of target area knowledge 501 Data An example of a database, wherein the database 601 an example of a database selection rule

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-3-94374 (JP, A) JP-A-2-87274 (JP, A) JP-A-1-200434 (JP, A) JP-A-2- 62669 (JP, A) JP-A-2-183372 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06F 17/30 JICST file (JOIS)

Claims (2)

(57) [Claims] 1. An information search system having a plurality of databases, a search request analysis unit for analyzing a search request input from an information requester; target area knowledge describing knowledge for selecting a database; A search request processing unit that converts a result of analysis by the search request analysis unit with reference to the target area knowledge into a search request format in which a search element is framed; A database feature database describing the database, a database selection rule for selecting a database that meets a search request from the plurality of databases, and the search according to the database selection rule with reference to the target area knowledge and the database feature database. Search data to select a database that meets the required format A database selection unit, for the database selected by the search request database selection unit, create a search formula in the search format, search the target database,
An information search system comprising: a search formula generation unit that receives and outputs a search result from a database; and a search result presentation unit that presents the search result. 2. A search method according to claim 1, further comprising: a database search generation rule for searching the respective databases with respect to the plurality of databases; and the search formula generator, wherein the search corresponding to the database selected by the search database selector is performed. A search expression generation unit starting unit that outputs the search request format to an expression generation unit, sets the database search generation rule, and starts the search expression generation unit; and a plurality of search results output from the search expression generation unit. 2. The information search system according to claim 1, further comprising a search result evaluation unit for removing duplicate data.