JPH11250073A - Method and device for retrieving semantic hierarchy of plural databases and medium for storing program therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To dissolve differences in data structures, addresses, data expression systems and classification levels of respective data bases in an environment where a plurality of database are connected to a network and to retrieve related data from the database against a keyboard inquiry. SOLUTION: In this hierarchical retrieval method, data structures, addresses of different databases, data expression systems, ranges of the values of data and classification levels of data are set in a table. A keyword is hierarchically retrieved so that it is traced from a route. The table is searched at the time of inquiry of database retrieval. Thus, differences are dissolved and the database is retrieved.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for retrieving a semantic hierarchy of a plurality of databases and a storage medium storing a retrieving program for a semantic hierarchy of a plurality of databases, and more particularly to an environment in which a plurality of databases are connected to a network. Even if the data structure, location, data expression format, data value range, data classification level, etc. of each database are different, resolve the differences and easily define the keyword tree structure data for the data value, A multi-database semantic hierarchy search method and apparatus and a multi-database semantic hierarchy search program are stored which search for keywords hierarchically so as to trace from the root, and which can search related data from the database in response to the keyword inquiry. It relates to a storage medium.

[0002]

2. Description of the Related Art In general, a large number of database systems have been introduced and used by companies for various tasks such as customer management and sales management. Initially, a database was used to streamline operations for simple data management purposes, such as storing, changing, and using customer and sales information. However, with the opening of networks in recent years, there is an increasing need to connect data stored so far to open networks for effective use. On the other hand, the use of databases is increasing in order to obtain information on management strategy, such as seeking information such as what is the best seller of products by region.

[0003] When a database is used to obtain management strategic information, there is a case where information from one database alone is not enough. Need to be integrated and used. However, since each database is operated independently, information of the same content is expressed in different expression formats (15,000 yen and $ 1).
5,000, etc.) or managing different value ranges (Kanto and Kansai data, etc.) with the same type of information,
Different classification levels for the same target data (daily sales,
Monthly sales, etc.).

[0004] On the other hand, in recent years, electronic commerce in which consumers and companies conduct various transactions such as sales and purchase of commodities using a computer network has been spreading. Electronic commerce is growing rapidly in recent years, and it is important to respond flexibly and quickly to market trends, such as the number of products and pricing, in order to increase sales while keeping rival companies down. Becomes Against this background, in electronic commerce, there has been an active movement to increase the flexibility of product information by dividing the product information embedded in the conventional homepage into a database to be managed and linking the WWW with the database. It is. Also, in the case of searching for product information, a hierarchical search is provided on the homepage of a selling company with many handled products to search for desired product information by tracing a keyword (PC → hardware → peripheral device, etc.). Account for the majority. In the hierarchical search, it is possible to easily determine whether the system includes the data required by the user by indicating the keyword of the data held by the system, and to narrow down the search range by following the keyword. Therefore, the demand for hierarchical search from the user to the database is increasing.

First, a method of integrating a plurality of databases has been studied and developed so far as a mechanism for integrating and using a plurality of databases. A multi-database system is a system that allows users to search distributed databases without having to be aware of the expression format and structure of individual information by predefining the virtual structure of the database used by the user, called a global schema. is there.

Next, a method for implementing a hierarchical search for one database is described. To provide a user with a hierarchical search for one database, it is necessary to hold hierarchical information by some means. As means for realizing conventional hierarchical information management, any of the following methods is used. (1) When using an application program:
The application manages the hierarchical structure information of the database data. The application manages the information of the hierarchical structure and the query sentence at the time of data search for the hierarchy completely independently of the database.

(2) In the case of realizing with a database: The hierarchical structure information of the data of the database is managed in the database. For example, first layer information, second layer information, third layer information
Manage hierarchical information, etc. in a database. The application does not need to be aware of hierarchical information. It is possible to realize a hierarchical search for a plurality of databases by combining these methods for implementing a hierarchical search of one database with the multi-database technology.

[0008]

However, in the above-mentioned conventional method, when an existing database that has been operated completely independently is newly added to the system, it is necessary to absorb the difference in the hierarchical structure information somewhere. Occurs, and it becomes necessary to review the database or the application.
When creating the hierarchical structure information, the administrator who manages and manages multiple databases needs to be aware of the data representation format, data value range, and data classification level. Difficult.

In the conventional methods, first, a global schema needs to be defined in order to integrate a plurality of databases, and for that purpose, it is necessary to know the data structure, location, data representation format, and the like of all the databases. In order to perform a hierarchical search on a database, it is necessary to investigate the value range of the database and the data classification level, define the hierarchical data of the keyword, and define the correspondence between the database and the hierarchical data of the keyword. There is. However, in order to perform these tasks, it is necessary to be familiar with the contents of all the databases, and it is actually very difficult. Therefore, there is a need for a mechanism that can eliminate these differences as much as possible without knowing the database structure, location, data expression format, data value range, and data classification level other than the own database.

Further, in order to integrate the data classification of the independently generated database, the fineness of the classification (daily and monthly,
Etc.) and differences in ideas (hardware / software and Wi
ndows / Mac, etc.)
It is necessary to design a common classification method for values in all databases and define the correspondence with each data. However, to perform this operation, it is necessary to reconsider the logical design of the multi-database and the design of the application depending on the level of the classification difference, which is very difficult. Therefore, there is a need for a mechanism that can generate hierarchical structure information of values for all databases without being conscious of differences in the level of fineness and ideas of classification other than the own database as much as possible.

The present invention has been made in view of the above points, and in an environment in which a plurality of databases are connected to a network, the data structure, location, data expression format, data classification level, and the like of each database are different. Also, it is possible to easily define the tree structure data of the keyword for the data value by eliminating the difference, search the keyword hierarchically so as to follow from the root, and search the database for the relevant data in response to the keyword inquiry. It is an object of the present invention to provide a method and an apparatus for searching a multiple database semantic hierarchy and a storage medium storing a multiple database semantic hierarchy search program.

[0012]

FIG. 1 is a diagram for explaining the principle of the present invention. According to the present invention (claim 1), in an environment in which a plurality of databases are connected to a network, even if the data structure, location, data expression format, data value range, data classification level, and the like of each database are different. Multi-database semantic hierarchy that resolves the difference, defines keyword tree structure data for data values, searches keywords hierarchically so as to follow from the root, and searches the database for related data in response to keyword queries In the search method, the names of the columns of each database and the ranges that are the ranges of the values are set in the column table, and before executing the search, the domain that is the expression format of the values of the columns of each database is set as the local domain. Set in the domain table, each domain having the same meaning as a domain group Global domain, which is typically used as a representative in each domain group, is set in the domain group table, a conversion function that performs conversion between domains is set in the conversion function table, and individual values of the global domain are set. Define keywords expressed in simple terms, define a meaningful hierarchical relationship between the keywords, set a keyword tree configured as tree-structured data in a keyword tree table, and, for each keyword tree, A table setting step (step 1) of setting a keyword tree search item, which is a search item used when the search item is omitted, in the keyword tree search item table;
A keyword search process (step 2) in which a user searches a target keyword by sequentially traversing a keyword tree upward and downward one step from the root, and a query sentence including a keyword, a search item, and a search condition obtained from the user. A database search step (step 3) of acquiring and searching the table set in the table setting step, converting the query sentence based on the obtained information, and searching the database.

According to the present invention (claim 2), in a database search process, a domain group of all keywords under a specified keyword in a keyword tree is added to a keyword of a query sentence including a user's keyword, a search item, and a search condition. A search is performed from the keyword tree table, a domain included in the domain group is searched from the local domain table, and a column range list that is a set of a combination of a column and a range as the domain is searched and generated from the column table. For each combination of the column range lists, if a predetermined condition is satisfied and the domain group of the column matches the domain group of one of the keywords under the designated keyword, the designated keyword is assigned to each column under the designated keyword and Domains in the range of each column Expand the value, to generate a first search condition connected by "OR" condition, the generated first search condition and the search condition specified by a user connected by "AND" condition,
If a second search condition is generated and the user does not specify a search item, a keyword tree search item of the specified keyword is searched from the keyword tree search item table, and the searched search item or the user By combining the search item specified in the above and the generated second search condition, a query sentence for searching the database is generated, and the search item of the query sentence for searching the database and the search data of the search condition are A local domain corresponding to the stored database and the search data is searched from a local domain table, and a global domain is searched from a domain group table based on a domain group including the local domain. If your local domain is different, A conversion function for converting the global domain to the local domain is obtained from the conversion function table, and the search items and search conditions are translated into a query suitable for each database using the conversion function, and each database is translated by the translated query. Search for.

According to the present invention (claim 3), when the first search condition is generated, as a predetermined condition, "a column having any one of the keywords under the specified keyword as a range" and "directly above the specified keyword" It is assumed that the column corresponds to one of a column having a range from any keyword from the keyword to the root, and a column having no range but matching a domain group of any keyword under the designated keyword.

FIG. 2 is a block diagram showing the principle of the present invention. According to the present invention (claim 4), in an environment in which a plurality of databases 40 are connected to a network, the data structure, location, data expression format, data value range, data classification level, and the like of each database 40 are different. Even if the difference is resolved, the tree structure data of the keyword for the data value is defined, the keyword is searched hierarchically so as to follow from the root, and the related data is searched from the database 40 in response to the keyword inquiry. A semantic hierarchical search device, which manages a column name of a column of each database 40 and a range which is a range of its value, and a column table 11 of each database 40 before executing a search. The local domain table 12 for managing a certain domain as a local domain and each domain having the same Set-in as a domain group,
A domain group table 13 for managing a global domain that is typically used as a representative in each domain group, a conversion function table 14 for managing a conversion function for performing conversion between domains, and individual values of the global domain. A keyword tree table 15 that defines keywords expressed in various terms, defines a semantic hierarchical relationship between the keywords, and manages a keyword tree configured as tree-structured data. A storage unit 10 having a keyword tree search item table 16 for managing a keyword tree search item, which is a search item used when the search item is omitted, and the user moves the keyword tree up or down one level from the root. Key to search for the desired keyword by tracing to And a query sentence including a keyword, a search item, and a search condition acquired from the user, search the storage unit 10, convert the query sentence based on the acquired information, and And a database search means 20 for searching for.

According to the present invention (claim 5), in the database search means 20, the domain group of all the keywords under the specified keyword in the keyword tree is compared with the keyword of the query sentence including the user's keyword, the search item, and the search condition. To Keyword Tree Table 1
5, a domain range included in the domain group is searched from the local domain table 12, and a column range list, which is a set of a combination of a column and a range as the domain, is searched from the column table 11 to generate a column range list. For each combination of the generation unit 21 and the column range list generated by the column range list generation unit 21, a predetermined condition is satisfied, and the domain group of the column matches the domain group of one of the keywords under the designated keyword. In this case, a keyword expanding unit 22 that expands a designated keyword for each column into domain values under the designated keyword and within a range of each column, and generates a first search condition connected by an “OR” condition; The first search condition generated by the means 22 and the search condition specified by the user are "AND" conditions A search condition combining means 23 for connecting and generating a second search condition; and a search item for searching a keyword tree search item table 16 for a keyword tree search item of a specified keyword when the user does not specify a search item. Search means 24 and search item search means 24
And a query sentence generating means 25 for combining the search item searched by the search item or the search item specified by the user with the query sentence and the generated second search condition to generate a query sentence for searching the database 40. And a database 40 storing search data of search items and search conditions of the query generated by the query generation unit 25.
A domain searching means for searching a local domain corresponding to the search data from the local domain table 12 and searching for a global domain from the domain group table 13 based on a domain group including the local domain; 2
If the global domain and the local domain searched in step 6 are different, a conversion function search unit 27 that obtains a conversion function for converting the global domain into a local domain from the conversion function table 14, and a conversion function search unit 27
Using the conversion function obtained in the above, the search item and the search condition are translated into a query sentence suitable for each database 40,
Search means 28 for searching each database 40 with the translated query sentence.

According to the present invention (claim 6), the predetermined condition used in the keyword expanding means 22 is "a column having any keyword under the specified keyword as a range",
Either a column whose range is any keyword from the keyword directly above the specified keyword to the root, or a column where the range is not set but matches the domain group of any keyword under the specified keyword Shall be applicable.

According to the present invention (claim 7), at least a managed column table 11, a local domain table 12,
Domain group table 13, conversion function table 1
4, a table setting means for setting the keyword tree table 15 and the keyword tree search item table 16 is included. According to the present invention (claim 8), in an environment in which a plurality of databases are connected to a network, even if the data structure, location, data expression format, data value range, data classification level, and the like of each database are different. Multi-database semantic hierarchy that resolves the difference, defines keyword tree structure data for data values, searches keywords hierarchically so as to follow from the root, and searches the database for related data in response to keyword queries A multi-data semantic hierarchical search program installed in a search device, comprising: a process of setting a column name and a range of a value of a column of each database in a column table; and executing each database before performing a search. Local domain table with the domain that is the format of the column value as the local domain The process of setting, the process of setting each domain having the same meaning as a domain group, the process of setting a global domain that is typically used as a representative in each domain group in the domain group table, and the conversion between domains A process for setting a conversion function in the conversion function table;
A process of defining keywords expressed in simple terms for individual values of the global domain, defining a semantic hierarchical relationship between the keywords, and setting a keyword tree configured as tree-structured data in a keyword tree table; A table setting process comprising, for each keyword tree, a process of setting a keyword tree search item, which is a search item used when a search item is omitted in a user inquiry, in a keyword tree search item table; Tree 1 from root
A keyword search process for searching for a target keyword by sequentially traversing up and down by rows, and a query sentence including a keyword, a search item, and a search condition obtained from a user are obtained, and the table set in the table setting process is obtained. It comprises a database search process for searching, converting the query sentence based on the acquired information, and searching the database.

According to the present invention (claim 9), in a database search process, a domain group of all keywords under a specified keyword in a keyword tree is added to a keyword of a query sentence including a user's keyword, a search item, and a search condition. A process of searching from a keyword tree table, a process of searching for a domain included in the domain group from a local domain table, and a process of searching a column table for a column range list that is a set of combinations of columns and ranges that are domains. For each combination of the process and the generated column range list, if the specified condition is satisfied and the domain group of the column matches the domain group of one of the keywords under the specified keyword, the specified keyword is specified for each column. The designated key A process of generating a first search condition connected to an “OR” condition by expanding to a value of a domain under a mode and within a range of each column, and generating the first search condition and a search condition specified by a user. Connect with "AND" condition,
A process of generating a second search condition, a process of searching a keyword tree search item table for a keyword tree search item of a specified keyword when the user does not specify a search item, a process of searching for the searched search item or A process of generating a query sentence for searching a database by combining a search item specified by a user in a query sentence and a generated second search condition; and a search item of a query sentence for searching a database. A database storing search data of search conditions and a local domain corresponding to the search data are searched from a local domain table, and a global domain is searched from a domain group table based on a domain group including the local domain. Process and the searched When the global domain is different from the local domain, a process of obtaining a conversion function for converting the global domain into the local domain from the conversion function table, and using the conversion function, the search item and the search condition are converted into a query sentence suitable for each database. The process includes a process of translating and a process of searching each database with the translated query.

According to the present invention (claim 10), as the predetermined conditions used in the process of generating the first search condition, "a column having any one of the keywords under the specified keyword as a range" and "directly above the specified keyword" Column that ranges from any keyword from the keyword to the root ”or“ a column that has no range but matches the domain group of any keyword under the specified keyword ” .

As described above, the user traverses the keyword tree one step at a time from the root upward and downward, and searches for a target keyword. Further, for a keyword of a query sentence including a user's keyword, a search item, and a search condition, a domain group of all keywords under the specified keyword in the keyword tree is searched from the keyword tree table. A domain included in the searched domain group is searched from the local domain table, and a column range list, which is a set of a combination of a column and a range as the searched domain, is searched and generated from the column table. For each combination of the generated column range list, if a predetermined condition is satisfied and the domain group of the column matches the domain group of any of the keywords under the specified keyword, the specified keyword is assigned to each column under the specified keyword. The search condition is expanded to domain values within the range of each column, and a search condition connected by an “OR” condition is generated. This search condition and the search condition specified by the user are referred to as “A
Connection is made under the “ND” condition, and a search condition is generated. At this time,
If the user does not specify a search item, the keyword tree search item of the specified keyword is searched from the keyword tree search item table, and the searched search item or the search item specified in the user query statement and the generated search condition To generate a query sentence.

A database storing search data of the generated query sentence and search data of search conditions and a local domain corresponding to the search data are searched from a local domain table, and the global domain includes the local domain. Search from the domain group table based on the domain group to be set.
If the searched global domain is different from the local domain, a conversion function for converting the global domain to the local domain is searched from the conversion function table,
The search item and the search condition are translated into a query sentence suitable for each database by using the searched conversion function, and each database is searched by the translated query sentence.

Thus, in an environment in which a plurality of databases are connected to a network, even if the data structure, location, data expression format, data value range, data classification level, and the like of each database are different, the differences are different. To easily define the tree structure data of the keyword for the data value, search the keyword hierarchically so as to follow from the root, and freely search the database for the related data in response to the keyword inquiry .

[0024]

FIG. 3 shows the configuration of a multiple database semantic hierarchical search apparatus according to the present invention. A plurality of database systems 200 having a database management unit 210 and a database 220 are connected to the multiple database semantic hierarchy search device 100 shown in FIG.
00, the user's terminal device 300 is connected.

Multiple Database Semantic Hierarchical Search Device 10
0 is a user interface unit 110, a keyword inquiry unit 120, a syntax analysis unit 130, an inquiry processing unit 140, a metadata storage unit 150, a metadata management unit 1
60 and a database access unit 170. The user interface unit 110 transmits and receives a keyword inquiry sentence, a database inquiry sentence, and a search result to and from a user.

The keyword inquiry section 120 processes a keyword inquiry sentence from the user and returns a list of applicable keywords. The syntax analyzer 130 analyzes the syntax of the query sentence from the user. The query processing unit 140 includes a query processing control unit 141, a query conversion unit 142, a conversion function library 143, and a search result processing unit 144. Inquiry processing control unit 14
1 controls the overall flow of processing for the analyzed query sentence. The inquiry conversion unit 142 converts an inquiry. The conversion function library 143 is a group of functions for performing data conversion between domains. The search result processing unit 144 converts the domain of the searched data.

The metadata storage section 150 is composed of a table shown in FIG. 4 for storing information on columns, domains, ranges, and keywords. The metadata storage unit 150 shown in FIG. 3 includes a column table 151, a domain table 152, a domain group table 153, a keyword tree table 155, and a keyword tree search item table 156.

The metadata management section 160 inputs metadata from an administrator. See Domain, Range,
For the keyword tree, there are eight settings: "Local domain settings", "Domain group settings", "Conversion function creation", "User domain settings", "Keyword settings", "Keyword tree settings" , "Keyword tree search item setting", and "range setting". Information such as the set domain, range, and keyword is stored in the metadata storage unit 150.

The database access unit 170 accesses a distributed database. The database access unit 170 is connected to one or more database systems 200 each including a database 220 and a database management unit 210 that manages the database 220. The database 220 is also called a local database.

FIG. 5 is a flowchart showing an outline of the operation of the present invention. In the configuration of FIG. 3, the system administrator sets a domain, a range, a keyword, and the like from the metadata management unit 160 (step 10), and stores the information of the set domain, range, keyword, and the like in the metadata storage unit 150. It is stored (Step 20) (Steps 10 and 20 are referred to as a preparation phase).

In this state, when the user inputs a keyword inquiry for acquiring a keyword from the application program 300, the keyword inquiry unit 120 via the user interface unit 110 of the multiple database semantic hierarchy search device 100 via the communication network 400. Is retrieved, and based on this keyword inquiry, a search is performed to retrieve the keyword from the keyword tree table by searching the metadata storage unit 150 (step 30). Also, when a query sentence including domain, range, and keyword tree information for searching the database from the user is input via the user interface unit 110, the syntax analysis unit 130 analyzes the query sentence and performs query processing control. The unit 141 searches the metadata storage unit 150 to obtain a domain, a range, and a keyword, and passes them to the inquiry conversion unit 142. The query conversion unit 142 acquires the conversion function from the conversion function library 143, converts the query sentence,
The database is searched by the database access unit 170 (step 40) (the processing of steps 30 and 40 is referred to as a search phase).

[0032]

Embodiments of the present invention will be described below with reference to the drawings. First, as a process before executing a search, a preparation phase for preparing a domain, a range, a definition of a keyword tree, and the like will be described. [Preparation Phase] In the preparation phase, the system administrator defines a domain, a range, and a keyword tree via the metadata management unit 160. FIG. 6 is a diagram for explaining domain management according to an embodiment of the present invention, and FIG.
FIG. 8 is a diagram illustrating a range according to an embodiment of the present invention, and FIG. 8 is a diagram illustrating a keyword tree according to an embodiment of the present invention.

As a domain which is a representation method of a data value of a column in a database, for example, as shown in FIG.
As a domain for indicating a price, in the case of expressing only numbers in yen units (15000), in the case of expressing numbers + “,” + “yen” (15,000 yen), expressing them in numbers + “1,000 yen” ( There are various expression methods such as the case of (15.0 thousand yen). In this way, the same meaning among various expressions is defined as a domain group, and only one domain is set as a domain that can be seen by the user, providing a single perspective to the user. can do.

In FIG. 6, the date is also listed as a domain group, and in this, year / month / day or month. Day. Year and the like are shown as various domains. Further, in FIG. 6, the user uses a number + “,” + “yen” for the price and year / month / day for the release date, and these are user domains. To set the range, which is the range of possible values for a column, define a domain for the column, define a domain group for that domain, define a global domain for that domain group, and set individual values for that global domain. Define a keyword expressed in plain terms, define a semantic hierarchical relationship to the keyword, define a keyword tree as tree-structured data, and include a range of possible values of a column on the keyword tree. A procedure for extracting one or more keywords is required.

[0035] If necessary, two or more keyword trees can be integrated into one keyword tree. Also, zero or more keyword trees can be defined in one system. For example, the product type code column of the product table handled in the database as shown in FIG. 7 is defined as a “product type classification code” domain, and the domain group of the domain is defined as a “product type classification” domain group.
The global domain of the domain group is defined as the "product type classification code" domain, and a keyword expressed in plain terms is defined for each value of the global domain, and a semantic upper limit relationship is defined for the keyword, and the tree structure data is defined. Figure 8 shows the result of defining a keyword tree as
The personal computer product type keyword tree shown in FIG.

The keyword tree is supplemented with a "product type large classification" domain group in order to form tree structure data. Each value of the product type code “11, 12,
"13, 21, 22" are "Company A's personal computer, B
The personal computer main unit, the company C personal computer main unit, the display, and the printer ”, and the column ranges defined so as to include these are as shown in FIG. Display; PC product type: Peripheral device / Printer;

The range setting “PC product type: PC body” is “PC product type: PC body / Company A PC, PC product type: PC body / B company PC, PC product type: PC body / C company PC” Individual designations indicate the same range. For domain, range, and keyword tree, eight settings, ie, setting of local domain, setting of domain group, creation of conversion function, setting of user domain, setting of keyword, setting of keyword tree, setting of keyword tree search item, setting of range are required. Yes, the system administrator has the metadata management unit 16
Using 0, information such as a domain, a range, and a keyword is set. These pieces of information are stored in the metadata storage unit 150.

FIG. 9 is a flowchart of the preparation phase according to one embodiment of the present invention. Step 101) Setting of a local domain: A "domain" which is a concept representing an expression form is set in a column of the local database 220. This is called a local domain. For example, the column "Price" in a database
(The type is char (16)), and the domain of the domain is set as “Price”. This “Price”
The meaning of the domain is not only a type of char (16), but also a more detailed expression form of “price”. For example, this Price: is in yen.

Numerical value, comma (,) and suffix "¥"
It consists of. It is an expression form. The definition of a domain for a column is held in a “column” table 151 as shown in FIG. Basically, this is done by the local database administrator. At that time, the already defined domain is referenced using the metadata management unit 160. If a domain that can be used as a domain for your database column is already defined, use it and do not create a new domain.

Step 102) Setting of a domain group: A group of domains that have different expressions but actually have the same meaning is called a domain group. The domains that fall into this group are all different representations of the same meaning. In the device 100 of the present invention, all the domains included in this domain group appear to be unified into one domain from the user.

The administrator of the local database 220
It determines which domain group the domain defined by itself belongs to, and performs processing to put it in that domain group. If no suitable domain group is found, create a new domain group. If the domain is already defined, create any domain group. If the domain is already defined, it does not need to be set because it is included in any domain group.

FIG. 11 shows an example of a domain table according to an embodiment of the present invention. One representative and standard domain is set in the domain group (this setting is basically performed by the system administrator of the apparatus 100). The domain is called a global domain. FIG. 12 shows an example of this global domain. Step 103) Creation of a transformation function: This is not necessary if a transformation function has already been defined. Create a conversion function between the global domain and the local domain using a general programming language. Compile your function into a form that can be dynamically linked.

The relationship between the conversion function and the domain is registered using the metadata management unit 160. The correspondence between the conversion function and the domain and the location of the conversion function can be found in the conversion function library 14
3, which is stored in the conversion function table therein. FIG. 13 is an example of a conversion function table according to an embodiment of the present invention. The conversion function creates a bidirectional function such as global domain → local domain, local domain → global domain. However, depending on the conversion, it may not be possible to create a function in both directions. Is generally not possible). In that case, create a function in only one direction. The reason for creating a bidirectional function is to perform conversion by combining functions when there is no function to directly convert.

Step 104) Setting of user domain: A domain which is actually visible to the user is selected from the domain group (for each user). This is called a user domain. When the user makes an inquiry to the apparatus 100, even if there are items implemented by various domains, the expressions are all unified to the user domain. Further, by changing the user domain, it is possible to change the appearance of data from the user. FIG. 14 shows an example of the domain group user table.

Step 105) Definition of Keyword: A keyword expressed in simple terms for each value of the global domain is defined. The keywords represent the contents of the data contained in the database, and when the user performs a semantic hierarchical search of the database, the keywords are to be selected. Therefore, terms that can be easily and uniquely identified are set. This setting is basically performed by the system administrator of the apparatus 100.

Step 106) Setting a keyword tree: A meaningful hierarchical relationship is defined for keywords, and a keyword tree is set as tree structure data. The keyword tree is a target for a user to search for a keyword for obtaining necessary information from a database. From the viewpoint of data management, it is necessary to construct a tree by classification according to a general concept of the user. .

The system administrator of the apparatus 100 determines which keyword tree contains the keyword defined by the user, and adds the keyword to the keyword tree table 155. If no suitable keyword tree is found, a new keyword tree is created. If necessary, two or more keywords can be integrated into one keyword. Further, zero or more keyword trees can be set in one system. FIG. 15 is an example of a keyword tree table according to an embodiment of the present invention.

Step 107) Setting of a keyword tree search item: A search item to be used by default is set in advance for each keyword tree. The keyword tree search item is used as the user's designated search item only when the search item is omitted in the user's inquiry. This setting is basically performed by the system administrator of the apparatus 100. FIG. 16 is an example of a keyword tree search item table according to one embodiment of the present invention.

Step 108) Range setting: To set a range, which is a possible range of values in the column of each database, one or more keywords including a range of possible values on the keyword tree are extracted. It is used to narrow down an inquiry destination database for an inquiry sentence including a user keyword, a search item, and a search condition.

Basically, the range is set by the administrator of the local database. The setting of the range is as shown in Fig. 1.
As shown at 0, it is held in the “column” table 151.
Next, the search phase will be described. FIG. 17 is a flowchart of the search phase according to one embodiment of the present invention. [Search Phase] The search phase executes a two-step process of a keyword inquiry phase for inquiring about a keyword, and a database inquiry phase for inquiring the database using an inquiry sentence including a keyword, a search item, and a search condition. The user generally calls and uses the functions of the apparatus 100 from an application program in the form of a library or the like.

First, the user interface unit 110 inquires a keyword from the user's terminal device 300,
Alternatively, when any of a query sentence including a keyword, a search item, and a search condition is obtained, it is determined which query is to be sent. Moves to step 205. In the database inquiry phase after step 205, a search is performed as follows using the information of the domain, range, and keyword tree. An inquiry from a user including a keyword, a search item, and a search condition is created in a language obtained by adding a KEYWORD statement to SQL, which is a search language of the relational database.

Step 201) Keyword inquiry phase First, the keyword search phase will be described. A search is performed using the keyword tree table 155. Inquiries from users by keyword search include "acquisition of keyword directly above specified keyword",
Two types of operations of “acquisition of a keyword immediately below a specified keyword” are provided.

Step 202) Acquisition of the keyword immediately above the designated keyword: This operation is received by the keyword inquiry unit 120, and the upper keyword of the keyword is acquired by searching the “keyword tree” table 155 of the metadata storage unit 150. Step 203) Acquisition of keyword immediately below the designated keyword: This operation is performed by keyword inquiry unit 1
20, a keyword having the specified keyword as a higher-level keyword is obtained by searching the “keyword tree” table of the metadata storage unit 150.

Step 204) The keywords acquired in steps 202 and 203 are returned to the user's terminal device. Thus, the user obtains a target keyword by interactively following the keyword tree up or down by combining the two types of operations. Step 205) Generation of a column range list: The query sentence is analyzed by the syntax analysis unit 130, keywords included in the query are extracted, and domain groups under the keywords are identified in the keyword tree table 15 shown in FIG.
Find out by searching for 5. Domain table 1 shown in FIG. 11 shows the domains included in the domain group.
Know by searching 55. FIG. 1 shows a column range list, which is a set of combinations of the domain, column, and range.
It is known by searching the “column” table 151 shown in FIG.

Step 206) Expand keywords into search conditions: The query processing control unit 141 determines, for each combination in the column range list, "a column having any one of the keywords under the specified keyword as a range" and "directly above the specified keyword." Column that ranges from any keyword from the keyword to the root), or a column that does not have a range but matches the domain group of any keyword under the specified keyword. It is known by searching the keyword tree table 155 shown in FIG. The keywords are expanded into keywords that are domain groups of keywords within the range of the column and under the designated keyword, and a search condition connected by an “OR” condition is generated.

Step 207) Combine search conditions: The query processing control unit 141 compares the search condition generated in the above-described step 206 with the search condition specified by the user by “AN”.
Connection is made under the condition "D", and a search condition is generated. This processing is not performed when the user has not specified the search condition. Step 208) Complementation of search item: When the user has not specified a search item, the search item of the keyword tree specified by the user is known by searching the keyword tree search item table shown in FIG. The obtained search item is used in subsequent processing as a search item specified by the user. This processing is not performed when the user specifies a search item.

Step 209) Generation of a query with keyword expansion: A query sentence composed of only the search item and the search condition by combining the search condition generated in step 208 and the search item generated in step 207 Generate Step 210) Query database determination: The database including the tables and columns included in the query generated in step 209 is known by searching the “column” table 151 shown in FIG.

Step 211) Search for the domain of the column included in the query: The domain of the column included in the query generated in step 209 is known by searching the "column" table 151 shown in FIG. Step 212) Search for user domain: By searching the "domain group user" table 153 shown in FIG.
Search for a user domain.

Step 213) Search for conversion function: Next, the query conversion unit 142 converts the conversion function between the user domain and the domain obtained in step 210 into the search function shown in FIG.
The conversion function table of the conversion function library 143 shown in FIG. If a conversion function that can be directly applied to A → B is not found, a search is made to see if conversion is possible in combination with A → C and C → B. This is a search problem on a graph in which a domain is a node and the existence of a transformation function between domains is an edge, and can be easily realized using an algorithm studied in graph theory.

Step 214) Translation and execution of the query: The query converter 142 converts the value included in the query from the user domain to the local domain in the following manner, thereby converting the value included in the query. Then, when the data is converted, the data access unit 170 accesses the database 220 of each database management system 200 using the search items, search conditions, and the like as parameters.

The search result processing unit 144 converts the value of the searched result from the local domain to the user domain. The user accesses a plurality of databases, and the result is the sum of the search results for each database. Next, the above operation will be described using a specific example.
18 to 20 show examples of a sample database according to one embodiment of the present invention.

In the following description, the domain, range,
For metadata such as keyword trees, see Figure 10
It is based on the values shown in FIG. The database A shown in FIG. 18 has the following format. Table "Handling products""Handlingproducts" column = (product type, product name, store name, price) Expression of "product type" = (A company computer, B company computer, C company computer, display, printer) "Product" Expression of “name” = (arbitrary character string indicating name of product) Expression of “price” = (15,000 yen) Database B shown in FIG. 19 has the following format.

Table “Handling Products” Column of “Handling Products” = (Product Type, Product Name, Store Name, Price) Expression of “Product Type” = (PC, Peripheral Equipment, Parts) Expression of “Product Name” (Any character string indicating the product type classification and the product name is connected using double-byte space characters.) Expression of “price” = (15.0 thousand yen) The database C shown in FIG. 20 has the following format.

Table “Handling Products” Column of “Handling Products” = (Product Type, Product Name, Store Name, Price) Expression of “Product Type” = (Display, Printer, Digital Camera) Expression of “Product Name” = (Arbitrary character string indicating the name of the product) Expression of “price” = (15,000 yen) The differences between these three databases A, B, and C are as follows. -Database A and Database C are classified as "product type" (Company A's PC, Company B's PC, Company C's PC, display, printer) and are classified as Database B's "Product Type"
Classification (computer body, peripherals, parts) is fine.

A product type classification is added to the product name in the database B. -While database A and database C use the format of "15,000 yen" for the expression of "price", database B uses the format of "15.0 thousand yen" for the expression of "price" I have. First, as a preparation phase, as described above, a local domain setting, a domain group setting, a conversion function creation, a user domain setting, a keyword setting, a keyword tree setting, a keyword tree search item setting,
The range setting is performed as follows.

(1) Setting of local domain (step 101): The administrators of the databases A, B, and C each set the local domain for their own database column. The domain of “product type” in the database A is an expression in words of the product type classification, and is “PCGoodsKindWord”.

The domain of the “product name” in the database A is just the product name, and is assumed to be “PCGoodsNameWord”. The "price" domain of database A is expressed in the form of "price number +", "+" yen ".
“PriceYen”. The domain of the “product type” in the database B is an expression in words of a large classification of the product type,
“PCGoodsKindClass”.

The domain of the “product name” in the database B is an expression combining the product type classification and the product name,
odsNameKind & Word ”.“ Price ”in database B
Is expressed in the form of “price number +“. ”+“ 1,000 yen ”” and is “PriceKYen”. The domain of the “product type” in the database C is an expression in words of the product type classification, and is “PCGoodsKindWord”.

The domain of the “product name” in the database C is just the product name, and is assumed to be “PCGoodsNameWord”. The "price" domain in the database C is expressed in the form of "price number +", "+" yen ".
“PriceYen”. For the columns other than the above, we do not need to specify the expression method and do not set the local domain.

(2) Setting of domain group (step 102): Here, as a domain group, PCGoodsKind (the domain included is PCGoodsKindWor
d, PCGoodsKindCode, Global domain is PCGoodsK
indWord) ・ PCGoodsClass (included domain is PCGoodsClassWo
rd, PCGoodsClassCode, global domain is PCGoodsC
lassWord), PCGoodsName (included domain is PCGoodsNameWor
d, PCGoodsNameKind & Word, Global domain is PCGo
odsNameWord), ・ Price (included domains are PriceYen, PriceKYen,
Set three global domains: PriceYen).

(3) Creation of a conversion function (step 10)
3): The following conversion function is defined for each domain group. Domain group: PCGoodsKind Conversion function: PCGoodsKindWord2PCGoodsKindCode () One-to-one word indicating product category classification to product category classification code
Perform the conversion. Conversion function: PCGoodsKindCode2PCGoodsKindWord () One-to-one conversion of product classification codes to words that represent product classifications
Perform the conversion. Domain group: PCGoodsClass Conversion function: PCGoodsClassWord2PCGoodsClassCode () The word that represents the product category major classification is 1 for the product category major classification code.
Perform one-to-one conversion. Conversion function: PCGoodsClassCode2PCGoodsClassWord () Converts the product type major classification code to a word that represents the product type major classification.
Perform one-to-one conversion. Domain group: PCGoodsName Conversion function: PCGoodsNameWord2PCGoodsNameKind & Word () Adds product classification type and double-byte space characters to the product name. Conversion function: PCGoodsNameKind & Word2PCGoodsNameWord () Deletes from the beginning of the specified character string to double-byte space characters. Domain group: Price Conversion function: PriceYen2PriceKYen () Divides the numerical value by 1000 and replaces "yen" with "1000 yen". Conversion function: PriceKYen2PriceYen () Multiplies the numerical value by 1000, and replaces “1,000 yen” with “yen”.

(4) User domain setting (step 1
04): In this example, it is assumed that the user domain is equal to the global domain (this is a case where the user user1 in FIG. 14 uses it). That is, (PCGoodKind user domain) = PCGoodsKindWord (PCGoodClass user domain) = PCGoodsKindWord (PcgoodsName user domain) = PCGoodsNameWord (Price user domain) = PriceYen.

(5) Definition of Keyword (Step 10)
5): In this example, "PCGoodsClass domain group"
And the global domain value of “PCGoodKind domain group” will be used as keywords.
In other words, "PCGoodsClass Domain Group" is: "PC" is "PC", "Peripheral" is "Peripheral", and "Parts" is "Parts". "PCGoodsKind Domain Group" is: "Company A PC" is "Company A PC", "PC B" is "Company B PC", and "PC C" is "Company C PC"-"Display" is "Display"-"Printer" is "Printer"-"Digital camera" is "Digital camera"-"CPU" is "CPU"-"Memory" is "Memory"- “Motherboard” is “Motherboard”, ・ “Video card” is “Video card”.

(6) Keyword Tree Setting (Step 106): FIG. 8 shows the concept of a personal computer product type keyword tree in which the keywords defined in the above (5) are defined as a tree structure database by defining a meaningful hierarchical relationship. . A table for managing the information of the keyword tree is a “keyword tree” table 155 as shown in FIG.

(7) Setting of Keyword Tree Search Items (Step 107): Three “tree name”, “store name”, and “price” are set as keyword tree search items for the personal computer product type keyword tree. (8) Range setting (step 108): The following ranges are set for each column having the domain of the domain group included in the keyword tree. "Product type" in the "Products" table of Database A
Column: ・ PC product type / PC main unit ・ PC product type / peripheral device / display ・ PC product type / peripheral device / printer “Product type” in “Handling product” table of database B
Column: ・ PC product type “Product type” in “Handling product” table of database C
Column: ・ For illustration purposes, no range is set.

Next, the keyword inquiry phase of the search phase will be described. The user traces the keyword tree from the root in the order of “root” → “PC product type” → “peripheral device” to acquire the keyword “peripheral device” (step 201 to step 204).

Finally, the database inquiry phase of the search phase will be described. The user's inquiry is converted into an SQL-like sentence as follows. Inquiry: SELECT No specification WHERE Price <70,000 yen KEYWORD “PC product type / peripheral device” (1) Generation of column range list (step 205): Domain of all keywords under the keyword “PC product type / peripheral device” The group is searched from the keyword tree table 155. First, "PC product type / peripheral device" domain group "PC
GoodsClass ”is searched. Next, "PCGoodsKind" which is a domain group of "display", "printer", and "digital camera" having the keyword "personal computer product type / peripheral device" as a higher keyword is searched.
After that, "PC product type / peripheral device / display", "PC product type / peripheral device / printer",
A search is performed for a keyword having any one of “PC product type / peripheral device / digital camera” as a higher keyword, but the search is terminated because there is no keyword. Thus, it is understood that the domain groups of all keywords under the keyword “PC product type / peripheral device” are as follows.

PCGoodsClass PCGoodKind Next, domains having “PCGoodsClass” or “PCGoodsKind” in the domain group are stored in the domain table 152.
Search from. As a result, the following is searched.・ PCGoodsKindWord ・ PCGoodsKindCode ・ PCGoodsClassWord ・ PCGoodsClassCode And “PCGoodsKindWord”, “PCGoodsKindCode”
"," PCGoodsClassWord "or" PCGoodsClassCod
A column and range list having a domain “e” is searched from the column table 151. As a result, the following column range list is obtained.

(Column Range 1) Column Name: Product Type, Table Name: Product, Database Name: Database A Range: PC Product Type / PC Body PC Product Type / Peripheral Equipment / Display PC Product Type / Peripheral Device / Printer (Column Range 2) Column Name: Product Type, Table Name: Product, Database Name: Database B Range: PC Product Type (Column Range 3) Column Name: Product Type, Table Name: Product, Database Name : Database C range: No setting (2) Expand keywords into search conditions (Step 20)
6): For each combination of the column range list generated in (1), “a column having any keyword under the specified keyword as a range” and “a keyword from the keyword immediately above the specified keyword to the root” Column that is a range ”or“ a column that has no range but matches the domain group of one of the keywords under the specified keyword ”, and the domain group of the column is one of the keywords under the specified keyword If it matches, the specified keyword is expanded for each column under the specified keyword into domain values within the range of each column, and a search condition connected by an "OR" condition is generated.

(Column range 1) The ranges "PC product type / peripheral device / display" and "PC product type /
Since the “peripheral device / printer” includes the keyword “personal computer product type / peripheral device”, this column range is determined to be “a column having any keyword under the specified keyword as a range”. Also, the column domain "PCGoodsKindWor
d) from the column table and find the domain "PCGoodsK
The domain group “PCGoodsKind” of “indWord” is searched from the domain table 152. The domain group “PCGoodsKind” matches the domain group “PCGoodsKind” under the keyword obtained in the above (1). Therefore, the specified keyword "PC product type / peripheral device"
Under the range, "PC product type / peripheral device / display" and "PC product type / peripheral device / printer"
In other words, the values are expanded to the values “display” and “printer” of the domain group “PCGoodsKind” in “PC product type / peripheral device / display” and “PC product type / peripheral device / printer”. The result is as follows.

(Product Type = “Display” OR Product Type = “Printer”) (Column Range 2) Since the keyword “PC product type / peripheral device” includes the range “PC product type”,
This column range is determined as "a column in which any keyword from the keyword immediately above the designated keyword to the root is set as a range". Also, the column domain "PCGoodsClassWo"
rd ”from the column table and find the domain“ PCGoodsC
Search "lassWord" domain group "PCGoodsClass" from the domain table. Domain group "PCGood
“sClass” matches the domain group “PCGoodsClass” under the keyword obtained in (1) above. Therefore, the domain group “PCGoodsClas” in the range “PC product type” under the specified keyword “PC product type / peripheral device”, that is, “PC product type / peripheral device”
s "to" peripheral equipment ".

As a result, the following is obtained. (Product type = "peripheral device") (Column range 3) Since this column range has no range, a domain group in the column is searched. That is, the domain “PCGoodsKindWord” of the column is searched from the column table 151, and the domain group “PCGoodsKind” of the domain “PCGoodsKindWord” is searched from the domain table 152. The domain group “PCGoodsKind” is a domain group (PC) under the keyword determined in (1) above.
GoodsKind, PCGoodsClass), it is determined that this column range is “a column that has no range but matches the domain group of any keyword under the specified keyword”. Therefore, the domain group “PC” under the specified keyword “PC product type / peripheral device”
Expands the value of "GoodsClass" to "display", "printer" and "digital camera". The result is: (Product type = “Display” OR Product type = “Printer” OR Product type = “Digital camera”) (3) Combining search conditions (Step 207): For each database, the search conditions generated in (2) above and the user The result of integrating the search conditions specified by is as follows.

(Column range 1) Column name: product type, table name: handled product, database name: database A (product type = “display” OR “product type =
“Printer”) AND price <70,000 yen (column range 2) Column name: product type, table name: product, database name: database B (product type = “peripheral device”) AND price <70,
000 yen (column range 3) column name: product type, table name: product, database name: database C (product type = “display” OR product type = “printer” OR product type = “digital camera”) A
ND price <70,000 yen (4) Completion of search item (step 208): Since the user did not specify a search item, search for a search item with the keyword tree name "PC product type" from the keyword tree search item table. . The results are as follows.

“Product Name, Store Name, Price” (5) Generation of Inquiry Expanding Keyword Specification (Step 209): Search Condition Generated in (3) and Search Item Generated in (4) for Each Database Will result in the following query: (Column range 1) column name: product type, table name: product name, database name: database A SELECT product name, store name, price WHERE (product name = “display” OR product type = “printer”) AND price <70, 000
Yen (column range 2) Column name: product type, table name: product, database name: database B SELECT Product name, store name, price WHERE (product type = "peripheral device") AND price <70,000 yen (column range 3) Column name: product type, table name: product, database name: database C SELECT Product name, store name, price WHERE (product type = “display” OR product type = “printer” OR product type = “digital camera”) AND price <70,000 yen (6) Determination of inquiry database (step 21)
0): The "column" table is searched, and a table including columns of product name, store name, and price is searched from the column table, and the following combinations are found.

Database A: Table name: Product line-up Database B: Table name: Product line-up Database C: Table name: Product line-up (7) Search for domain of column included in query (step 21)
1): (8) Search for user domains (step 212): Search for user domains of groups that include each domain. (9) Search for a conversion function (step 213): The process searches for a “user domain → local domain conversion function” for “value included in conditional expression”.

Database B: Price → PriceYen2PricekYen () (10) Value conversion (step 214): The result of executing the conversion function and converting the value is shown below. Database B: PriceYen2PriceKYen (70,000
The result of (yen) is 70.0 thousand yen.

(11) Translation of Inquiry (Step 2)
14): Translate the inquiry. Database A: SELECT Product Name, Price FROM Product WHERE (Product Type = “Display” OR Product Type = “Printer” AND Price <70,000 Database B: SELECT Product Name, Price FROM Product WHERE (Product Type = "Peripheral equipment") AND Price <70.0 thousand yen Database C: SELECT Product name, price FROM Handling product WHERE (Product type = "Display" OR Product type = "Printer" OR Product type = "Digital camera") AND Price <70,000 yen (12) Executing an inquiry to each database (step 214): Execute an inquiry to each database A, B, C.

Database A: Product name: DSP-
1, Store name: Shop A, Price: 50,000 yen ・ Database B: Product name: Digital camera, DC-
1, Store name: Shop E, price 56.0 thousand yen ・ Database C: Product name: DSP-1, Store name: Shop F, Price: 49,8000 yen ・ Database C: Product name: DC-1, Store name: Shop F , Price 56,800 yen (13) Conversion of search result (step 214): Database B: The price is converted since the user domain and the local domain are different.

Database B: Product name PCGoodsNameKin
d & Word2PCGoodsNameWork (Digital camera DC-1)
Is "DC-1". Price PriceKYen2PriceY
The result of en (56.0 thousand yen) is “56,000 yen”. (14) Generation of sum of search results (step 214): Sum of search results. The final search result is: Product name: DSP-1, Store name: Shop A, Price: 50,000
Yen Product name: DC-1, Store name: Shop E, Price: 56,000 yen Product name: DSP-1, Store name: Shop F, Price: 49,800
Yen Product name: DC-1, Store name: Shop F, Price: 56,800 yen Also, without being limited to the above embodiment, a search device shown in FIG. Connected disk devices,
The present invention can be easily realized by storing it in a portable storage medium such as a floppy disk, a CD-ROM, or the like, and installing it when implementing the present invention.

The present invention is not limited to the above embodiments, but can be variously modified and applied within the scope of the claims.

[0091]

As described above, according to the present invention, in an environment in which a plurality of databases are connected to a network, the data structure, location, data expression format, data value range, and data classification of each database are provided. Even if the levels are different, it is easy to define the tree structure data of the keyword for the data value by resolving the difference, search the keyword hierarchically so as to follow from the root, and retrieve the relevant data in response to the keyword inquiry. Can be freely searched from the database.

Further, it is not necessary to define a global schema, and it is possible to eliminate a difference in data classification. In addition, in the existing multi-database and hierarchical search technology, differences cannot be eliminated without knowing the data structure, location, data representation format, data value range, and data classification level of all databases. Thus, in the present invention, the local database administrator has a database data structure, location, data representation format, data value range, data classification level, and global domain which is a standard representation format. It is only necessary to be aware of the keyword tree, and it is not necessary to be aware of the data structure, location, data expression format, data value range, and data classification level of other databases.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the principle of the present invention.

FIG. 2 is a principle configuration diagram of the present invention.

FIG. 3 is a configuration diagram of a multiple database semantic hierarchical search device of the present invention.

FIG. 4 is a diagram illustrating a configuration of a table of a metadata storage unit according to the present invention.

FIG. 5 is a flowchart showing an outline of the operation of the present invention.

FIG. 6 is a diagram illustrating domain management according to an embodiment of the present invention.

FIG. 7 is a diagram for explaining a range according to an embodiment of the present invention.

FIG. 8 is a diagram illustrating a keyword tree according to an embodiment of the present invention.

FIG. 9 is a flowchart of a preparation phase according to an embodiment of the present invention.

FIG. 10 is an example of a column table according to an embodiment of the present invention.

FIG. 11 is an example of a domain table according to an embodiment of the present invention.

FIG. 12 is an example of a domain group table according to an embodiment of the present invention.

FIG. 13 is an example of a conversion function table according to an embodiment of the present invention.

FIG. 14 is an example of a domain group user table according to an embodiment of the present invention.

FIG. 15 is an example of a keyword tree table according to an embodiment of the present invention.

FIG. 16 is an example of a keyword tree search item table according to an embodiment of the present invention.

FIG. 17 is a flowchart of a search phase according to an embodiment of the present invention.

FIG. 18 shows a sample database A according to an embodiment of the present invention.
This is an example.

FIG. 19 is a sample database B according to one embodiment of the present invention.
This is an example.

FIG. 20 is a sample database C according to one embodiment of the present invention.
This is an example.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Storage means 11 Column table 12 Local domain table 13 Domain group table 14 Conversion function table 15 Keyword tree table 16 Keyword tree search item table 20 Database search means 21 Column range list generation means 22 Keyword expansion means 23 Search condition combination means 24 Search items Search means 25 query sentence generation means 26 domain search means 27 conversion function search means 28 search means 30 keyword search means 40 database 100 multiple database semantic hierarchy search apparatus 110 user interface section 120 keyword query section 130 syntax analysis section 140 query processing section 141 Query processing control unit 142 Query conversion unit 143 Conversion function library 144 Search result processing unit 150 Metadata Data storage unit 151 Column table 152 Domain table 153 Domain group table 155 Keyword tree table 156 Keyword tree search item table 160 Metadata management unit 170 Database access unit 200 Database system 210 Database management unit 220 Database 300 Terminal device 400 Communication network

Claims (10)

[Claims] In an environment in which a plurality of databases are connected to a network, a data structure of each database,
Even if the location, data expression format, data value range, data classification level, etc. are different, resolve the differences to define the tree structure data of the keyword for the data value, and hierarchically define the keyword so that it can be traced from the root In a multi-database semantic hierarchical search method that searches the database for related data in response to keyword inquiries, the name of each database column and the range of its value are set in the column table, and the search is performed. Before executing, set the domain that is the expression format of the column value of each database as a local domain in the local domain table, set each domain having the same meaning as a domain group, and represent each domain group as a representative. Global group domain table used as standard Setting, setting a conversion function for performing the conversion between the domains in a conversion function table, defining keywords expressed in plain terms in individual values of the global domain, and defining a semantic hierarchical relationship between the keywords. A keyword tree configured as tree-structured data is set in a keyword tree table. For each of the keyword trees, a keyword tree search item, which is a search item used when a search item is omitted by a user inquiry, is defined as a keyword. A table setting step of setting a tree search item table; a keyword search step of a user searching for a target keyword by sequentially traversing the keyword tree one step up or down from the root; a keyword obtained from the user; Obtain a query sentence consisting of search items and search conditions, and A method for retrieving a semantic hierarchy of a plurality of databases, comprising searching a table set in a table setting process, converting the query sentence based on the acquired information, and searching the database. 2. In the database search process, a domain group of all keywords under a specified keyword in the keyword tree is retrieved from the keyword tree table for the keyword of a query sentence including a user's keyword, a search item, and a search condition. Searching, searching the local domain table for domains included in the domain group, searching and generating from the column table a column range list that is a set of combinations of columns and ranges that are the domains. For each combination in the column range list,
When the predetermined condition is satisfied and the domain group of the column matches the domain group of any keyword under the designated keyword, the designated keyword is changed to the value of the domain under the designated keyword and within the range of each column for each column. Expand, generate a first search condition connected by an "OR" condition, connect the generated first search condition and a search condition specified by a user by an "AND" condition, and generate a second search condition If the user does not specify a search item, a keyword tree search item of the specified keyword is searched from the keyword tree search item table, and the searched search item or a search item specified by the user in a query sentence And the generated second search condition to generate a query sentence for searching a database, A database storing search data of a query sentence and search data of a search condition for searching for a query, and a local domain corresponding to the search data are searched from the local domain table, and a global domain is included in the local domain. Searching from the domain group table based on the domain group to be searched, and if the searched global domain and the local domain are different, obtain a conversion function for converting the local domain into a global domain from the conversion function table; 2. The multiple database semantic hierarchical search method according to claim 1, wherein the search item and the search condition are translated into a query sentence suitable for each database by using a conversion function, and each database is searched by the translated query sentence. 3. The method according to claim 1, wherein when the first search condition is generated, as the predetermined condition, “a column having a range of any keyword under the specified keyword”; A column whose range is any keyword "," A column whose range is not set but matches the domain group of any keyword under the specified keyword "
3. The method according to claim 2, wherein the method corresponds to any one of the following. 4. In an environment in which a plurality of databases are connected to a network, a data structure of each database,
Even if the location, data expression format, data value range, data classification level, etc. are different, resolve the differences to define the tree structure data of the keyword for the data value, and hierarchically define the keyword so that it can be traced from the root A multi-database semantic hierarchical search device that searches for relevant data from a database in response to a query for a keyword, and manages a column name of each database and a range that is a range of values thereof, a column table and Before executing the search, set the local domain table that manages the domain that is the expression format of the column value of each database as a local domain, and set each domain having the same meaning as a domain group, and represent each domain group as a representative. Domain group that manages global domains used as standard Table, a conversion function table for managing a conversion function for performing conversion between domains, and defining keywords expressed in simple terms in individual values of the global domain, and defining a semantic hierarchical relationship between the keywords. A keyword tree table that manages a keyword tree configured as tree-structured data, and manages, for each of the keyword trees, a keyword tree search item that is a search item used when a search item is omitted in a user inquiry Storage means having a keyword tree search item table for performing a search; a keyword search means for searching for a target keyword by a user sequentially traversing the keyword tree one step up or down from the root; and a keyword obtained from the user Get a query sentence consisting of And a database search unit that searches the storage unit, converts the query sentence based on the acquired information, and searches the database. 5. The database search means, for a keyword of a query sentence including a user's keyword, a search item, and a search condition, retrieves a domain group of all keywords under a specified keyword in the keyword tree from the keyword tree table. Explore,
A column range list generating unit that searches the local domain table for a domain included in the domain group, and searches and generates a column range list, which is a set of combinations of columns and ranges that are the domains, from the column table; For each combination of the column range list generated by the column range list generating unit, if a predetermined condition is satisfied and the domain group of the column matches the domain group of any of the keywords under the designated keyword, Keyword expansion means for expanding the specified keyword into domain values under the specified keyword and within the range of each column, and generating a first search condition connected by an "OR"condition; The first search condition and the search condition specified by the user are connected by an “AND” condition, and the second search condition is connected. Search condition combination means for generating a condition; search item search means for searching a keyword tree search item table for a specified keyword from the keyword tree search item table when the user does not specify a search item; Means for generating a query sentence for searching a database by combining the search item searched by the means or the search item specified by the user with the query sentence and the generated second search condition A database storing search data of the query sentence and search data of the search condition generated by the query sentence generating means and a local domain corresponding to the search data from the local domain table; To the domain group containing the local domain Domain search means for searching from the domain group table based on the domain group table, and a conversion function for converting the global domain into a local domain when the global domain searched by the domain search means is different from the local domain. Using a conversion function search unit obtained from the conversion function obtained by the conversion function search unit, a search item and a search condition are translated into a query sentence suitable for each database. 5. The multiple database semantic hierarchy search device according to claim 4, further comprising search means for searching each database. 6. The predetermined condition used in the keyword expanding means includes: “a column having a range of any one of the keywords under the specified keyword”; Column to be set as a range ”,“ Column that has no range but matches the domain group of one of the keywords under the specified keyword ”
6. The multiple database semantic hierarchical search device according to claim 5, wherein the device corresponds to any one of the following. 7. at least a column table to be managed;
5. The multiple database semantic hierarchy search device according to claim 4, further comprising table setting means for setting the local domain table, the domain group table, the conversion function table, the keyword tree table, and the keyword tree search item table. 8. In an environment in which a plurality of databases are connected to a network, a data structure of each database,
Even if the location, data expression format, data value range, data classification level, etc. are different, resolve the differences to define the tree structure data of the keyword for the data value, and hierarchically define the keyword so that it can be traced from the root Is a multi-data semantic hierarchy search program installed in a multi-database semantic hierarchy search device that searches for relevant data from a database in response to a query for a keyword. It has the same meaning as the process of setting the range that is the range in the column table and the process of setting the domain that is the expression format of the column value of each database as the local domain in the local domain table before executing the search. Set each domain as a domain group, and within each domain group A process of setting a global domain that is typically used as a representative in the domain group table, a process of setting a conversion function for performing conversion between domains in the conversion function table, and a process of setting individual values of the global domain in a simple manner. A process of defining keywords expressed in terms of terms, defining a semantic hierarchical relationship between the keywords, and setting a keyword tree configured as tree-structured data in a keyword tree table; A table setting process including a process of setting a keyword tree search item, which is a search item used when the search item is omitted in the keyword tree search item table, and a user moves the keyword tree one stage from the root. To go up and down one by one A keyword search process for searching for a target keyword, acquiring a query sentence including a keyword, a search item, and a search condition obtained from the user, searching the table set in the table setting process, and obtaining the obtained information. A storage medium storing a multi-database semantic hierarchy search program, comprising a database search process for converting the query sentence based on the database and searching the database. 9. The database search process includes, for a keyword in a query sentence including a user's keyword, a search item, and a search condition, a domain group of all keywords under a specified keyword in the keyword tree from the keyword tree table. A process of searching; a process of searching for a domain included in the domain group from the local domain table; and a process of searching for and generating a column range list, which is a set of combinations of columns and ranges, being the domain, from the column table. And For each combination of the generated column range list,
If the specified condition is satisfied and the domain group of the column matches the domain group of any keyword under the specified keyword, the specified keyword is expanded to the value of the domain under the specified keyword and within the range of each column for each column. Generating a first search condition connected by an “OR” condition; connecting the generated first search condition to a search condition specified by a user by an “AND” condition; A process of generating a keyword tree search item of a specified keyword from the keyword tree search item table when the user has not specified a search item; and a process of searching the searched search item or the user Query for combining the search item specified in step 2 with the generated second search condition and searching the database A process of generating a set sentence; searching a database storing search data of a query sentence and search data of a search condition for searching the database and a local domain corresponding to the search data from the local domain table; A process of searching the domain group table for a global domain based on a domain group including the local domain; and converting the global main to the local domain if the searched global domain and the local domain are different. A process of acquiring a function from the conversion function table; a process of using the conversion function to translate a search item and a search condition into a query sentence suitable for each database; Storage medium storing a plurality database semantic hierarchy search program as recited in claim 8, including a process to search over scan. 10. The predetermined condition used in the process of generating the first search condition includes: “a column having any keyword under the specified keyword as a range”; “from a keyword immediately above the specified keyword to a root”. Column with any of the keywords in the range "," column with no range set but matching the domain group of any of the keywords under the specified keyword "
10. A storage medium storing the multiple database semantic hierarchy search program according to claim 9, which corresponds to any one of the above.