WO2011074729A1

WO2011074729A1 - Ontology-based instance identification system and method

Info

Publication number: WO2011074729A1
Application number: PCT/KR2009/007593
Authority: WO
Inventors: 김평; 정한민; 이승우; 이미경; 김재한; 류범종
Original assignee: 한국과학기술정보연구원
Priority date: 2009-12-18
Filing date: 2009-12-18
Publication date: 2011-06-23
Also published as: KR20110069987A; KR101081870B1

Abstract

Disclosed is an ontology-based instance identification system and method. The present invention is advantageous of improving the data access through either information refinement or semantic inference by providing a uniformed resource identifier (URI) to the relational database management system (RDBMS) data, in a process of converting the RDBMS data into the instance of web ontology language (OWL) ontology.

Description

Ontology-based Instance Identification System and Its Method

The present invention relates to an ontology-based instance identification system and method thereof, and more particularly, in a process of converting data of a relational database management system (RDBMS) into an instance of a web ontology language (OWL) ontology. The present invention relates to an ontology-based instance identification system and method for assigning unique identifiers (URIs) to data of a database management system to express relationships between instances.

Today, as the type and number of information increases exponentially, the time required for researchers to acquire and analyze the required information is gradually increasing. However, existing information services such as keyword search are vertical for specific domains. There is a problem in that they cannot meet the demand for a more accurate search service through search or semantic processing, and user satisfaction is gradually lowered accordingly.

The semantic web technology for solving the above problems expresses the meaning and association of words using ontologies, and extends the semantic relations through inference and expresses them with various services, thereby improving accessibility and user satisfaction. Is also contributing to.

However, it is not easy to replace existing information services with semantic services because it is not easy to express and manage knowledge with the practical level and ontology of semantic web technology.

Therefore, many existing data are stored and managed in RDBMS, and in order to provide semantic services for them, it is necessary to convert DBMS data into instances of ontology.

In the conventional technology of converting RDBMS data into ontology, researches for automatically generating ontology for RDBMS schema or converting into ontology instance using DB schema and ontology schema mapping information have been conducted.

However, when the existing RDBMS data is not given a Uniform Resource Identifier (URI) or needs to be newly identified, it is difficult to simply switch to an ontology instance.

In the case of ontology instances, since URIs are assigned and the relationships between instances are expressed through URIs, it is necessary to newly assign URIs in the data transformation process. However, in the prior art, a method of newly granting and managing URIs through instance identification in the data transformation process is provided. Did not cover.

Therefore, in order to solve the above-mentioned problem, the present invention converts data of a relational database management system (RDBMS) into an instance of Web Ontology Language (OWL) ontology, and generates an ontology instance with an identifier to provide accurate information. It is to provide an ontology-based instance identification system and method for improving the accessibility of information through finding or semantic inference.

In order to achieve the above object, after ontology is registered by ontology register and ontology register which register ontology database of ontology database of class, attribute, and relationship between classes, ontology database, instance to be identified is entered. Is a class-instance connector that selects and connects a class to which a class is assigned, a reference and additional information obtainer for obtaining information necessary for instance identification, and information obtained to identify an instance. If the identification target instance is identified as the same instance as the existing instance by the controller, the identification identifier is assigned the same identifier assigned to the existing instance among the instances of the ontology registered in the ontology DB. Be And an ontology-based instance identification system, comprising: an instance identifier granter for assigning a unique identifier

The ontology register of the ontology-based instance identification system registers an ontology consisting of class, attribute, and relationship information between classes in the ontology DB of the database.When the ontology is registered, the class-instance connector receives an input of an instance to be identified. Selecting and attaching a class to be allocated to the instance; Acquiring information necessary for identification of an instance to be identified by the reference and additional information obtaining unit; identifying an instance of an input identification object among instances of a registered ontology, and identifying an instance of an instance of a registered ontology If the instance is not identified, the instance identifier granter provides a unique identifier; to provide an ontology-based instance identification method comprising a.

The ontology-based instance identification system and method thereof according to the present invention generate an ontology instance with identifiers when converting data from a relational database management system (RDBMS) into an instance of a web ontology language (OWL) ontology. It is effective to improve the accessibility of information through finding information or semantic reasoning.

1 is a diagram illustrating an OntoURI system for automatically generating RDBMS data as an ontology instance given a URI by using a mapping rule of a DB schema and an ontology schema and an identification rule defined to identify an instance.

2 is a diagram illustrating a data conversion process using an OntoURI system.

3 is a diagram illustrating an ontology-based instance identification system according to the present invention;

4 is a diagram illustrating interdependence between instances in an ontology-based instance identification system according to the present invention;

And,

5 is a flowchart illustrating an instance identification method by an ontology-based instance identification system according to the present invention.

The terms or words used in this specification and claims are not to be construed as limiting in their usual or dictionary meanings, and the inventors may properly define the concept of terms in order to best explain their invention in the best way possible. It should be interpreted as meaning and concept corresponding to the technical idea of the present invention.

Hereinafter, an ontology-based instance identification system and method thereof according to the present invention will be described in detail with reference to the accompanying drawings.

First, before describing the ontology-based instance identification system and method thereof according to the present invention, an OntoURI system for automatically generating RDBMS data corresponding to a system higher than the present invention as an ontology instance given a URI will be briefly described.

FIG. 1 is a diagram illustrating an OntoURI system for automatically generating RDBMS data as an ontology instance given a URI by using a mapping rule of a DB schema and an ontology schema and an identification rule defined to identify an instance.

As shown in FIG. 1, the OntoURI system supports a user interface for creating a DBMS- ontology mapping rule and transmits the generated rule to the server. The rule editor 110 configures a server environment for linking external services and resources. Rules generated by the server management unit 120, the monitoring unit 130 to determine whether the DBMS data changes, the statistical unit 140 for providing statistical information on the progress or the created instance, the rule editing unit 110 Rule management unit 150 for managing the data, instance identification unit 160 for instance identification by applying to the instance identification rule, the scheduler 170 for registering and performing a schedule for the data conversion operation, mapping rules and DB monitoring results Instance generation unit 180 to create an instance using the DB, DB monitoring unit 190 and ontology to grasp the change of the DBMS to be converted It comprises a management information model (200) for storing and managing the schema instance.

The data conversion process by the OntoURI system having the above-described components will be briefly described with reference to FIG.

2 is a diagram illustrating a data conversion process using an OntoURI system.

As shown in FIG. 2, the data conversion process by the OntoURI system is composed of seven steps.

The seven-step data conversion process mentioned above is to set up the environment for inference engine, search engine, and other external services through the server environment setup, and the JDBC environment for DBMS interworking, and ontology schema registration. Step 2 specifying the URI prefix and format for the server, step 3 registering the connection information of the target DB, step 4 creating the mapping rule and registering it with the URI server using the mapping rule editor, and authority data required for the instance identification rule. And five steps of creating identification rules, six steps of performing data transformation operations through the scheduler, and seven steps of delivering the final generated URI to the inference engine and the search engine through a web service.

In particular, in the OntoURI system which automatically generates the above-described RDBMS data into an ontology instance given a URI, the present invention normalizes data by using authority data and identification rules in the process of converting DB data into an ontology instance, and normalized data. In order to assign and manage a unique identifier, the above-described instance identification unit 600 and steps 4 and 5 in the data conversion process will be described.

3 is a diagram illustrating an ontology-based instance identification system according to the present invention.

As shown in FIG. 3, the ontology-based instance identification system according to the present invention includes an ontology register 300, a class-instance connector 400, a controller 500, a reference and additional information obtainer 600, and an instance identifier adder. 700 and the database unit 800.

The ontology register 300 registers an ontology consisting of classes, attributes, and relationship information between classes in the ontology DB 810 of the database unit 800.

As a specific embodiment, in the case of the thesis class, since the meta information of the article is related to the publisher, the author, and the subject of the article, the ontology registrar 300 may refer to the author or subject of the publisher referred to to assign a unique identifier to a single article. Register information and related information.

After the ontology is registered by the ontology registerer 300, the class-instance connector 400 selects and connects a class to which an instance is assigned when an instance to be identified is input.

The reference and additional information obtainer 600 acquires information necessary for instance identification when the instance to be identified and the class of the instance are determined.

The control unit 500 has the information obtained by the reference and additional information acquirer 600, and identifies whether the input identification target instance exists among the ontology registered in the ontology DB 810.

When the instance to be identified is present and identified in the ontology DB 810 by the control unit 500, the instance identifier assigning unit 700 gives the same identifier as the instance identified in the ontology DB 810, and identifies the object. If the instance does not exist because it does not exist in the ontology DB 810, a unique identifier is assigned to the input identification target instance.

The ontology-based instance identification system according to the present invention having the above-described configuration will be described in more detail how instance identification is performed in the process of converting thesis DB data into an ontology instance.

The authority data referenced by the reference and additional information obtainer 600 is used to obtain representative values of the name organization publishers for the names actually represented in the academic information metadata as described in the table below.

Table 1

Representative value	Actual value	type	class
IBM	International Business Machines Corporation	full name	institution
Microsoft	MS	abbreviation	institution
ACADEMIC PRESS, INC.	Academic Press	expression	publication

When the authority data is described in detail based on the contents of the above table, the reference and additional information acquirer 600 is the international name corresponding to the institution name with the full name as the actual value of the instance to be identified. When the authority data indicating that "Corporation" is represented by the representative value "MS" is obtained, the control unit 500 has the information obtained by the reference and additional information acquirer 600, and represents the actual value "International Business Machines Corporation" and the representative. The value "IBM" is judged to be the same value and the identification of the entered instance is performed.

Similarly, the control unit 500 recognizes the actual value "MS" as an abbreviation of "MicroSoft" which is a specific institution and identifies the instance to be identified.

If there is a representative value corresponding to the actual value, the actual value is treated as an instance with the same identifier as the identifier already assigned to the actual value, and if there is no representative value corresponding to the actual value, the actual value is identified as an instance. The rule is applied to the existing instance to which the identifier is assigned, and if it is determined to be a new instance, the new identifier is given to the instance of ontology.

An academic paper ontology is composed of a class including a paper, an institution, a topic, an author, a source, and local information. The information instance identification is identified in order from an instance of a class having low interdependence between classes, as shown in FIG.

4 is a diagram illustrating interdependence between classes in an ontology-based instance identification system according to the present invention.

In FIG. 4, the dependency between classes is indicated by a solid line, and information used for identification is indicated by a dotted line.

Only authority data is used for geographical name identification. When a particular city is identified, a unique identifier assigned to the city is selected by referring to the city name and the country name.

That is, when a unique identifier is assigned to the representative value "Seoul", "SEOUL" corresponding to the actual value is selected and assigned to the "Seoul" based on authority data.

The present invention focuses on the author name identification, and various qualities are used to identify the author's name, such as the article written by the author, the institution of the author, the e-mail address of the author, the subject of the author's article, and the author's co-author information.

TABLE 2

class	endowment	division	Same	Degree	source	weight
person		order				One
person	name	pivot	exact	single	Ontouri
person	hasInstiution	feature	exact	multiple	Ontouri	0.7
person	hasCoauthor	feature	exact	number	Ontouri	0.3
person	threshold					0.6

[Table 2] above is part of the instance identification rule for identifying the author's name. The instance identification rule is as follows.

The class is the class to be identified, the qualities are the attributes used for identification, the classification is the description of how to apply the qualities to the instance, the match is the value match or the number of values matched, the order is the relation order, and the source is the attribute The sources and weights from which values can be obtained are the importance of each property.

On the other hand, the application order belonging to the division is an instance identification order, which prioritizes instance identification of a class with low dependence, and a reference value is a candidate group to which the same URI can be assigned when the values of specified qualities are the same. A criterion for classification, a feature, is a criterion to give the same URI as the existing instance when the attribute and threshold final weight of the instance exceeds the threshold value.

In addition, an match belonging to a match is a case where values match exactly, and a number is a number of values that match exactly.

And, the single belonging to the order is a case where only one attribute value exists, and the multiple is a case where two or more attribute values exist.

When an administrator defines a rule for identifying instances on the OntoURI server and assigns a weight to each rule, OntoURI applies the rules defined to identify the instance during instance identification, and then adds a threshold with the sum of the final weights. The same URI or different URI will be allocated depending on whether or not it is exceeded.

The instance identification rule and the weight of each rule may be selected in consideration of the property value of the corresponding domain and the importance of each property.

In addition, when the information associated with the identifier of the existing DB is converted through the URI mapping information given through the identifier of the existing DB and OntoURI, the URI mapped with the DB identifier is used.

An instance identification method by an ontology-based instance identification system having the above components will be described.

The ontology register 300 of the ontology-based instance identification system registers an ontology, which consists of class, attribute, and relationship information between classes, in the ontology DB 810 of the database unit 800 (S100).

After the ontology is registered by the ontology register 300, when an instance to be identified is input, the class-instance connector 400 selects and connects a class to be assigned to the instance (S200).

When the selection of the class associated with the instance to be identified in step 'S200' is completed, the reference and additional information obtainer 600 performs a step of obtaining information necessary for instance identification (S300).

When the selection of the class associated with the instance to be identified in steps S200 and S300 is completed, and information necessary for identification is obtained, the controller 500 selects an input identification instance among the registered ontology instances. Identifies a step (S400).

If the identification target instance input by the control unit 500 is not identified in step S400, the instance granting unit 700 performs a step of assigning a unique identifier (S500) and is input from an instance of a registered ontology. If the identified instance is identified, a step of assigning the same identifier as that of the identified instance is performed.

Although the present invention has been described in detail only with respect to specific examples described in the embodiments, it is apparent that various modifications and variations can be made by those skilled in the art within the scope of the present invention, and such variations and modifications also belong to the appended claims. Of course it is.

The present invention relates to an ontology-based instance identification system and a method thereof, wherein an instance to be identified in the process of converting data of a relational database management system (RDBMS) into an instance of a web ontology language (OWL) ontology Is entered, it identifies the existence of the instance to be identified to the instance of ontology already registered, and if the identified instance is identified, assigns the same identifier as the identified instance, and if it is not identified, assigns a new unique identifier. It is effective to find the correct information or improve the accessibility of information through semantic reasoning.

Claims

An ontology register 300 for registering an ontology, which consists of classes, attributes, and relationship information between classes, in the ontology DB 810 of the database unit 800;

A class-instance connector 400 that selects and connects a class to which an instance is assigned when an ontology is registered by the ontology register 300 and an instance to be identified is input;

A reference and additional information obtainer 600 for acquiring information necessary for identifying the instance when the instance to be identified and the class of the instance are determined;

The control unit 500 having the information obtained to identify the instance obtained by the reference and additional information obtainer 600, and identifying the input identification target instance among the instances of the ontology registered in the ontology DB 810. ; And

An instance identifier granter 700 for assigning the same identifier as the instance identified in the ontology DB 810 when the target object to be identified by the controller 500 is identified, and for assigning a unique identifier if not identified. Ontology-based instance identification system, characterized in that.
The method of claim 1,

The reference and additional information obtainer 600 is

An ontology-based instance identification system, characterized by obtaining authority data comprising representative values of an instance registered in a class of an ontology, and actual values having different meanings but identical representations.
The method of claim 1,

The control unit (500) is an ontology-based instance identification system, characterized in that from among the instances of the ontology registered in the ontology DB (810), in order from the low interdependence instance in order.
The method of claim 1,

The instance identifier providing unit 700 is

An ontology-based instance identification system according to claim 1, wherein a weighted instance identification rule is applied to identify the instance, and a unique identifier is assigned according to whether the sum of the final weights exceeds a specified threshold.
(a) registering the ontology DB (810) of the database unit (800) by the ontology register (300) of the ontology-based instance identification system;

(b) when the ontology is registered in step (a), the class-instance connector 400 receives an input of an instance to be identified, selecting a class to be allocated to the instance, and connecting the ontology;

(c) obtaining, by the reference and additional information obtainer 600, information necessary for identification of the instance to be identified;

(d) after the step (c), the controller 500 identifies the input identification target instance among the registered ontology instances; And

(e) if the instance to be identified is not identified to the instance of the ontology registered in the step (d), the instance identifier assigning unit 700 gives a unique identifier; the ontology, characterized in that it comprises a How to identify the base instance.
The method of claim 5,

Step (e) is

and (e`) when identifying an instance to be identified to an instance of the registered ontology, assigning the same identifier as the identified instance.
The method of claim 5,

In step (d),

The control unit 500

Ontology-based instance identification method, characterized in that to identify among the instances of the ontology registered in the ontology DB (810) in order from the instance with low interdependence.
The method of claim 5,

In the step (e),

The instance identifier providing unit 700 is

Applying a weighted instance identification rule to identify the instance, and assigning a unique identifier according to whether the sum of the final weights exceeds a specified threshold.