JP2010086437A - Retrieval system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a mapping system for a retrieval system, which has an information structure and a function for associating existing data with ontology to construct ontology and associating the ontology with existing database, thereby easily expanding the coverage of objects to be processed based on ontology. <P>SOLUTION: The retrieval system includes the mapping system comprising: a metadata creating means for creating metadata structured from the existing database; a projection creating means for creating a projection for extracting specific information from the metadata structured by the metadata creating means; a first mapping means for mapping the projection created by the projection creating means to a class of an information ontology database; a second mapping means for mapping a projection item of the projection to a property of the class; and a third mapping means for mapping the value of an effective value list of the projection item to the value of an effective value list of the property. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a search system having a mapping system that has an information structure and function for associating existing data with an ontology, and constructs the ontology and associates it with a database of an existing information system.

  Patent Document 1 considers an ontology from a database and uses a single indexing system to extract information such as documents, images, and other portable multimedia information. Computer including one or more front-end computers and one or more computer nodes interconnected by a network to form a single search engine for obtaining objects processed by the ontology・ Disclose the database system. Further, in this Patent Document 1, each object is adapted to a specific ontology, and the query is one of the objects adapted to the specific ontology, which is used for searching for an object adapted to one or more target ontology. It is. The query from the user is sent to one of the front-end computers, which forwards the query to one of the computer nodes of the search engine of the present invention called the home node. The home node extracts features from this query according to the ontology. Then these features are hashed. Each hashed feature and target ontology list is sent to one node on the network. Each node on the network that receives the hashed feature uses the hashed feature of this query to perform a search against each corresponding portion of the database. The search result of the local database consists of this query, the ontology on which the object was processed, and the object identifier of the object located in the hashed equivalent feature in the other ontology. These other hashed features are transferred to each corresponding node as needed, and the process continues until the desired target ontology is reached. When the target ontology is reached, local database search results are collected by the home node. The result of this query computes each target ontology, and this process is repeated by the home node to refine the query results. That is, in the invention according to Patent Document 1, a plurality of features are extracted from a query according to the query ontology, each of the features is divided into feature fragments, and each of the feature fragments is hashed into a hashed feature fragment. Thus, an information retrieval system is provided for processing a query to retrieve information from a database based on words or based on something other than words.

  The content access device disclosed in Patent Document 2 refers to content data using an ontology that is similar to the ontology of the application when the ontology used in the content is different from the ontology used by the application that refers to the content. In order to be able to do so, if there is a means to access the ontology database that holds the ontology, and the two ontologies used by the content and the application are different, use the third ontology in the ontology database to The mapping is performed to adjust the correspondence relationship.

In the method of ontological integration of heterogeneous biological data disclosed in US Pat. No. 6,099,089, the location from which the data from the first data set is analyzed and the components of the first data set are mapped to the ontology is determined. decide. Data from the second data set is analyzed to determine where the components of the second data set map to the ontology. A subset of the first data set is identified and a subset of the second data set is identified. A first ontology item is identified by a component of a subset of the first data set. A subset of the second data set is statistically analyzed for mapping to the ontology, and a second ontology item is identified by a component of the subset of the second data set. An association of the first ontology item with the second ontology item can further be performed.
JP-T-2002-521753 JP 2001-75988 A JP 2006-120143 A

  In general, an information system is constructed to support individual business processes, and data is generated and stored with an information structure and grammar optimized for the business processes. However, in order to make the best use of the accumulated data assets, it is desirable that the data processed in each business process be generated and stored in a single standardized structure and grammar. , An information system in which data is generated and stored in different data structures and grammars with different cultures, languages, terms, and standards will be constructed at different locations, and effective use by exchanging data between different information systems will be a heavy burden Become.

  In addition, information ontology technology that can structure the concepts of objects in the target world and express them with semantics without duplication or deficiency can generate and store data based on a single standardized concept. Has been attracting attention.

  An information system that builds an ontology and generates data based on the ontology has a large data conversion load on an existing information system and an information system reconstruction load as is apparent from the above-described patent literature. In addition, only information systems constructed to support ontologies are within the scope of benefits.

  For this reason, the present invention has an information structure and function for associating existing data with an ontology, and by constructing the ontology and associating it with a database of an existing information system, the information processing target by the ontology can be easily obtained. The object is to provide a search system that can be expanded.

  The search system according to the present invention is a mechanism for combining an information ontology class that has been made into a knowledge as a specific concept and an existing database that has the same or equivalent information as the concept that has been made into knowledge in that class. By mapping information in an existing database with different data structure, grammar, and standards to each structure of the information ontology, the data in the existing database is made compatible with data standardized in the information ontology.

  Further, by mapping a plurality of existing databases to a single information ontology, the data of each existing database having different data structures, grammars, and standards can be interchanged through the information ontology.

  In the mapping system according to the present invention, the structure of the mapping between the information ontology and the existing database is based on the metadata of the existing database, the projection of the existing database approximate to the class definition is structurally defined, and each part of the projection and each part of the information ontology are It can be realized by associating.

  Therefore, the search system according to the present invention is constructed for the purpose of supporting a business process based on an information ontology that has a class that becomes knowledge based on the concept of an object and that is linked with one or more properties that specify the class. In a search system for performing a search from an existing database used by an existing information system that is in operation, a metadata creating unit that creates structured metadata from the existing database, and a structure created by the metadata creating unit Projection creating means for creating a projection for extracting specific information from the converted metadata; first mapping means for mapping the projection created by the projection creating means to a class of an information ontology database; and A state in which a class and a projection are mapped by one mapping means The second mapping means for mapping the projection item of the projection to the property of the class, and the projection item with the property having the valid value list and the projection item having the valid value list mapped by the second mapping means And a third mapping means for mapping the value of the valid value list to the value of the valid value list of the property.

  The metadata includes physical database information, information on data files stored in the database, information on connection between data files, information on data items in the data file, and information on valid data values of the data items. It is desirable.

  The projection creating means further comprises a unit database, and when the projection structure definition defines that the units are different with respect to the projection item of the data having the unit concept, the unit database has a compatibility method between the units. It is desirable to convert the unit by searching from.

  It is desirable to obtain information from an existing database via at least one projection mapped to the desired class by the mapping system by selecting a desired class from the information ontology.

  Therefore, according to the present invention, the information ontology database does not have the instance data of the class, and the data in the database processed by the existing information system is handled as the instance data. Therefore, a plurality of existing databases are used as they are. There is an effect that can be.

As shown in FIG. 1, the search system according to the present invention includes a mapping system 1 and search processing means 600 composed of steps 100 to 500. This mapping system 1 has a class that becomes knowledge based on the concept of an object, and is constructed and operated for the purpose of supporting a business process based on an information ontology in which one or more properties specifying the class are linked. In a search system for performing a search from an existing database used by an existing information system, metadata creating means 100 for creating metadata of the existing database, and metadata created by the metadata creating means 100 Projection creation means 200 for creating a projection for extracting specific information, first mapping means 300 for mapping the projection created by the projection creation means 100 to a class of the information ontology database, and this first mapping Class and projection are mapped by means 300 The second mapping means 400 for mapping the projection item of the projection to the property of the class, and the second mapping means 400 maps the property having the valid value list and the projection item having the valid value list. In this state, the third mapping means 500 is configured to map the value of the effective value list of the projection item to the value of the effective value list of the property.

  In the metadata creating means 100, the metadata is data related to data, and the metadata of the existing database is data related to data stored in the existing database. The metadata of an existing database includes physical database information, information on data files stored in the database, information on connection between data files, data file data item information, and data item valid data value information. is there. The metadata of these existing databases can be automatically generated from the catalog information held by the management system of the existing database. In addition, it is possible to supplement the missing information manually.

  Projection creation in the projection creation means 200 is shown in the flowchart of FIG. 2, for example. In creating a projection starting from step 200, one data file in an existing database is selected in step 201. In step 202, it is determined whether or not to further select a data file. When selecting (Y), the process returns to step 201 to select another data file, and it is determined not to select (N) in the selection in step 202. Until the data file is selected. As a result, one or more data files in the existing database are selected.

If it is determined in step 203 that the number of selected data files is 2 or more (Y), the process proceeds to step 204, where the combination conditions for the selected data files are set, and the projection item in step 205 is defined. If the number of selected data files is 1 (N), the process skips step 204 and proceeds to step 205 to define a projection item. This definition is
1. Define a projection item that directly references a single data item.
2. Define a projection item that computes and derives a single data item.
3. Define a projection item that calculates and derives multiple data items.
4). Define projection items that do not refer to data items and are dynamically derived from constants or environments.
That is.

  Specifically, as shown in FIG. 2, it is determined in step 206 whether or not the projection item is a simple reference to a data item. If it is a simple reference (Y), the data item is selected in step 207. The If it is not a simple reference (N), the process proceeds to step 208. If the projection item refers to a data item (Y), the process proceeds to step 209 to select the data item, and the selection is completed in step 210. Until then, the selection of the data item in step 209 is repeated, and then the process proceeds to step 211 to define an expression for deriving the projection item. Further, even when the projection item does not refer to the data item in the determination in step 208, the process proceeds to step 211 to define an expression for deriving the projection item. In step 207 and steps 211 to 212, the names of the projection items are defined.

Thereafter, in step 213, a semantic data type is set. There is a semantic data type applied to the projection item (the indented data side is treated as a subtype of the data type of the previous line).
・ Logical type ・ String type ・ Variable length character type ・ Fixed length character type ・ Non-quantity code type ・ Numeric type ・ Integer type ・ Quantity integer type ・ Quantity level integer type ・ Non-quantity integer type ・ Passing integer type ・ Real number type ・Quantity real type ・ Quantity level real type ・ Passing real number type ・ Date time type

In step 214, the unit is set. As this unit,
-Units of measure: meters, grams, liters, degrees Celsius. . .・ Currency units: Japanese yen, US dollar, Euro, Hong Kong dollar. . . Etc. ・ Year-of-year: Western calendar, Japanese calendar. . .・ Time zone: Japan Standard Time, US West Coast Time (Daylight Saving Time). . . Etc. Thereby, it is possible to detect that the unit is different between the property mapped in the future and the projection item. When the user or the computer system detects that the units are different, the user or the computer system can obtain a knowledge of performing unit conversion by searching the unit database for a compatibility method between the units.

  Further, in step 216, it is determined whether or not there is a valid value list for the projection item. If not (N), the process returns to step 220 from the control routine. If there is (Y), the process proceeds to step 217 to create an effective value list, an effective value is added to the effective value list in step 218, and step 218 continues until it is determined in step 219 that there is no effective value. If no more (N), return to the base routine in step 220.

  The project item definition work from step 205 is executed until there is no projection item defined in the determination in step 230. If it is determined that all projection items have been defined (N), the process proceeds to step 231 to project. The search condition for the record that falls within the range is defined (extraction condition setting). The search condition for the record that is the projection range is defined so as to provide exclusive information to other projections defined in the same existing database. If this exclusive projection definition is not guaranteed, certain data in the existing database is mapped to a plurality of information ontology classes, so that the concept represented by the information ontology becomes ambiguous. The projection creating means 200 ends in step 232 and returns to the main routine.

The first mapping unit 300 maps projections to information ontology database classes. However, projection and class mapping have the following restrictions.
• Projections cannot be mapped to multiple classes.
A class can be mapped to multiple projections.
A class cannot map multiple projections defined for the same existing database.
The mapping between information ontology classes and projections defined for any existing database is 1: 1.

  The second mapping unit 400 maps the projection item of the projection to the property of the class in a state where the class and the projection are mapped. Only those whose property data type and the semantic data type of the projection item match can be mapped. Only a project item having a valid value list can be mapped to a property having a valid value list.

  The third mapping unit 500 maps the value of the effective value list of the projection item to the value of the effective value list of the property in a state where the property having the effective value list and the projection item having the effective value list are mapped. It is something to be made.

  Information written in the language in the definition of the information ontology class, property, property valid value, etc., can be registered in any number of languages in the same language, and the same meaning for each language Is stored in a structure that can register an arbitrary number of different notations. This makes it possible to find the same information from "TV", "TV", "TV", "TV" in Japanese, "television" in English, "TV" in Chinese, etc. It is.

  The search processing means 600 is, for example, as shown in FIG. In the search process 600, first, in step 601, an information ontology is selected, and a predetermined class in the information ontology is selected. In step 603, the projection mapped to the selected class is acquired. In step 604, the structure definition of the projection is acquired. In step 605, the property / projection item of the class to be extracted is selected. In step 606, an extraction condition (search condition) is specified. In step 607, a search request for the existing database is generated. In step 608, the existing database is searched. In step 609, the search result data is acquired. In step 610, the units are changed and matched. In step 611, the search result data is returned. In step 612, the search processing operation is completed.

  From the above, when the mapping system according to the present invention is schematically described, a block configuration as shown in FIG. 5 is obtained. In this block configuration, the metadata 20 of the existing database extracted from the existing database includes at least database information, data file information, data file item information, and a data file item value list.

  A projective structure definition 30 exists at the conceptual upper part of the metadata 20 of the existing database. Projections are generated by this projection structure definition, specifically, data file selection, data file connection definition, data file item selection, data extraction definition, projection item expression definition, and projection item value list.

  This projection is mapped to a class included in the information ontology database 50 via a projection-information ontology mapping 40, the projection item is mapped to a class property, and the value of the projection item value list is the property value of the class. Mapped to list value.

  In FIG. 6, this will be described in more detail. Metadata 20 extracted from the existing database 10 includes database file information Tbl1, Tbl2, Tbl3, Tbl4 and database file item information C11, C12, C13, C21, C22, C23, C31, C32, C33, C41, C42, C43, and a database file item value list Lv.

  The database file information Tbl1, Tbl2, Tbl3, Tbl4 is selected as a data file through the data extraction condition 34 and the data file combination condition 35, and is generated into projections Prj1, Prj2, Prj3, and the database file item information C11-C43 is Data file items are selected and projection items Pjc1-1 to Pjc3-3 are generated. Further, the database file item value list Lv is generated as a projection item value list Pjv. These constitute the projective structure definition 30.

  The projections Prj1, Prj2, Prj3 are mapped to the classes Cls1, Cls2, Cls3 of the information ontology database 50 via the projection-information ontology mapping 40. At the same time, for example, projection items Pjc1-1 and Pjc2-1 are mapped to the class property Prp1-1 of the class Cls 1, and projection items Pjc1-2 and Pjc2-2 are mapped to the class property Prp1-2. Projection items Pjc1-3 and Pjc2-3 are mapped to the class block patty Prp2-1. The projection item value list Pjv is mapped to the class property value list Prpv.

  Thus, for example, when class Cls2 is selected, projections Prj1 and Prj2 are selected, Pjc1-1 to Pjc2-3 are selected as projection items, and database file information Tbl1 to Tbl4 are selected as metadata, respectively. Based on this, an item of an existing database is selected, and data is downloaded according to the extraction condition.

  An example of retrieving information in a specific existing database from the information ontology is shown below.

  As shown in FIG. 7 (a), for example, metadata db01 created based on an existing database has a product classification code and a product classification name as product classification information, and SKU, representative JAN code 8 as product information. It has 13 digits, representative JAN code, product name and classification, SKU, company code, purchase start, purchase end and purchase price as purchase information, company code and company as supplier company information Have a name. In addition, as shown in FIG. 7B, the metadata db02 has a department CD and a department name as product department information, a product name CD, a product name and department CD as product name information, and a product CD as sales product information. , Commodity CD country, commodity CD item, commodity CD check digit, commodity name, commodity receipt display name, commodity name CD and registration date, country CD and country name as EAN country CD information, business partner information It has CD and supplier name, has product CD, supplier company, start date and end date as purchase information, and has product CD, start date, end date and price as price information.

  Further, as shown in FIG. 8, the class of the information ontology database having the product name, product short name, product classification, product command destination company, product purchase start date, product purchase end date, product purchase price as properties ( : Product name) Projection: product name of db01 electrical product and projection: Product name of db02 electrical product is mapped, Projection name for product short name: Project name of product receipt display of db02 electrical appliance is mapped, For product classification, projection: db01 electrical product product classification name and projection: db02 electrical product department name is mapped, product supplier company projection: db01 electrical company name and projection: db02 electrical business partner Name is mapped, product purchase start date is projected: db01 electrical appliance purchase start and projection: db02 electrical product start date is mapped, product purchase end date Projection: db01 appliances purchase end and projection: db02 appliances end date is mapped, about the product purchase price projection: db01 purchase price of electrical appliances is mapped.

  In addition, as shown in FIG. 9, in this embodiment, projection: projection item of db01 appliance: product category name has a product category code and a product category name as a valid value list, and projection: projection item of db02 appliance : The department name is accompanied by a department CD and a department name as a valid value list, and these are mapped to class: appliance property: valid value list product classification and formal name.

  An example of searching for product information in an existing database from the information ontology with the above configuration is shown below. When a user obtains a list of audiovisual products and their purchase price from an existing database, the user scans and selects a class: appliance from the information ontology. As described above, the class: electrical appliance has property: product name, property: product short name, property: product classification, property: product purchase end date, and property: product purchase price. Also, class: electrification product is mapped to projection: db01 electrical appliance, projection: db02 electrical appliance. The user selects property: product name, property: product classification, property: product purchase price as items necessary for the list.

  Property: Product name, Property: Product classification, Projection: db01 electrical appliance, Projection: db02 It is mapped to the corresponding projection item of electrical appliance, but the purchase price is projected: Projection of db01 electrical appliance: Purchase Only the price is mapped. For this reason, projection: db01 electrical appliance is selected as the projection to be used. Property: product name is mapped to projection item: product name, property: product classification is mapped to projection item: product classification code, property: product purchase price is mapped to projection item: purchase price.

  Projection: db01 electrical appliance is structured from db01 metadata, Projection item: Product name is data file item: Product name, Projection item: Product classification code is Data file: Product classification item: In the product classification code, the projection item: purchase price is mapped to the data file: purchase item: purchase price. Mapping from this two-layer mapping to information ontology (class: electrical appliance) → existing database: db01 is performed. When the existing database: db01 is an RDBMS, the following SQL is obtained as a request from the ontology to the existing database from the metadata of the existing data: db01.

select Product classification. Product classification code as product classification,
Product. Product name as product name,
Purchase. Purchase price as Product purchase price
from Product Classification
left join product on (Product classification. Product classification code = Product. Classification)
left join purchase on (product.SKU = purchase.SKU);

  Since the user needs a list of audiovisual products, the user sets property: value of effective value list of product category, product category: 2 (audio video product) as a search condition. Property: Value of product category effective value list, Product category: 2 is mapped to Projection item: Value of product category effective value list, Product category code: DD200 (see FIG. 9). Thereby, the condition is added to the SQL as follows.

select Product classification. Product classification code as product classification,
Product. Product name as product name,
Purchase. Purchase price as Product purchase price
from Product Classification
left join product on (product classification, product classification code = product.classification)
left join purchase on (product.SKU = purchase.SKU);
when Product classification. Product classification code = 'DD200';
Necessary information can be obtained by executing this request on the existing database: db01.

It is the flowchart figure which showed the structure of the search system which comprises the mapping system which concerns on this invention. It is the flowchart figure which showed creation of the projection of the mapping system which concerns on this invention. It is the flowchart figure which showed the definition of the projection item of the mapping system which concerns on this invention. It is the flowchart figure which showed the search process of the search system which concerns on this invention. 1 is a structural block diagram of a mapping system according to the present invention. It is explanatory drawing which showed an example of the mapping system which concerns on this invention. (a), (b) is explanatory drawing which showed an example of the metadata of the mapping system based on the Example of this invention. It is explanatory drawing which showed an example of the mapping of the projection item and class which concern on the Example of this invention. It is explanatory drawing which showed an example of the mapping of the effective value list which concerns on the Example of this invention.

Explanation of symbols

100 metadata creation means 200 projection creation means 300 first mapping means 400 second mapping means 500 third mapping means 600 search processing means

Claims (4)

An existing information system constructed and operated for the purpose of supporting a business process based on an information ontology that has a class that becomes knowledge based on the concept of an object and that is linked with one or more properties that specify the class. In a search system that searches from an existing database in use,
Metadata creation means for creating structured metadata from the existing database;
Projection creating means for creating a projection for extracting specific information from the metadata structured by the metadata creating means;
First mapping means for mapping the projection created by the projection creation means to a class of an information ontology database;
A second mapping means for mapping a projection item of the projection to a property of the class in a state where the class and the projection are mapped by the first mapping means;
In the state where the property having the effective value list and the projection item having the effective value list are mapped by the second mapping means, the value of the effective value list of the projection item is mapped to the value of the effective value list of the property. A search system comprising a mapping system comprising three mapping means.   The metadata is physical database information, information on data files stored in the database, information on connection between data files, information on data items of data files, and information on valid data values of data items. The search system according to claim 1, wherein:   The projection creating means further comprises a unit database, and when the projection structure definition defines that the units are different with respect to the projection item of the data having the unit concept, the unit database has a compatibility method between the units. The search system according to claim 1 or 2, wherein unit conversion is performed by searching from (1).   4. Information from an existing database is obtained by at least one projection mapped to a desired class by the mapping system by selecting a desired class from the information ontology. The search system as described in any one.

Images