WO2013008979A1 - Entity-identifying device and method using multiple ontologies - Google Patents

Abstract

The present invention relates to an entity-identifying device and method using multiple ontologies, comprising: a triple storage unit for storing, in triplicate, entity identification results on each entity; a property information collecting unit for acquiring property information on a new entity identifier from a multiple ontology database when the new entity identifier is inputted; a first grouping unit for grouping the new entity identifier on an entity-type basis using property-expressing types from among the acquired property information; a second grouping unit for regrouping the entity identifiers on the basis of a representative property of the entity identifiers included in groups according to each entity type; and an entity identifier group determining unit for acquiring and comparing related property values for each group of entities grouped by the second grouping unit, dividing each of said groups into subgroups on the basis of compared results, and determining entity identifier groups. Thus, according to the present invention, property information on multiple ontologies included in linked data can be used to progressively identify entities of various types.

Description

Individual identification device and method using multiple ontology

The present invention relates to an apparatus and method for identifying an object using multiple ontologies. More particularly, when a new entity identifier is input, the attribute information of the new entity identifier is obtained from a multiple ontology database, and the type is obtained from the acquired attribute information. After grouping the new entity identifier based on the entity type using an attribute representing the entity type, regrouping the entity identifiers based on representative attributes of the entity identifiers included in the group according to each entity type, and regrouping the entity identifiers. An apparatus and method for identifying individuals using multiple ontology which obtains and compares association attribute values of entities of a group and determines an entity identifier group by dividing each group into subgroups based on the comparison result.

As the importance of information sharing and linking is highlighted, web services using ontology as a knowledge expression model are increasing. Recently, data model using ontology has been attracting attention in connection with research of semantic web.

Applications using the semantic web include automated communication between agents, web service automation, semantic-based search services in ubiquitous environments, and information retrieval from heterogeneous multimedia databases. However, these applications basically assume that all agents reference a common ontology. If these agents do not refer to a common ontology, such an application would be impossible. That is, since the semantic web is distributed and heterogeneous like the existing web, it is difficult to assume that all agents refer to a common ontology.

For example, if one agent refers to an ontology that describes the address of a person's home as "address", the other agent refers to an ontology that describes it as a "postal_address". Communication between them becomes impossible.

In addition, existing studies to ensure the consistency of URIs necessary for linkage and utilization of semantic web data have limited types of entity identification and do not effectively support real-time entity identification of ontology, which is gradually added. There are limitations such as using the error as it is.

The present invention has been made to solve the above problems, and an object of the present invention is to identify an object using a multiple ontology that can gradually identify various types of objects by using attribute information of multiple ontology included in Linked Data. And providing a method.

Another object of the present invention is to group the ontology into the same type and object by using the attribute information of multiple ontologies, and to improve the accuracy of the same object identification by identifying the object through the comparison of the main property of each group, as well as compared to the external service It is to provide an object identification device and method using multiple ontology that can support the incremental object identification function targeting only the URI added through.

According to an aspect of the present invention to achieve the above object, when the object identification result for each individual is stored in triple form triple storage unit, when the new object identifier is input, attribute information of the new object identifier from the multiple ontology database Acquisition attribute information collecting unit, a first grouping unit for grouping the new entity identifier based on the object type by using the attribute indicating the type from the obtained attribute information, representative of the entity identifiers included in the group according to each entity type A second grouping unit for regrouping the object identifiers based on the property, and obtains and compares the associated property value for the objects of each group grouped in the second grouping unit, and based on the comparison result Includes an object identifier group determination unit that divides into subgroups to determine an object identifier group. Provided is an individual identification device using multiple ontologies.

The entity identification device may further include a verification unit configured to repeat the process of determining the entity identifier group for the new entity identifier to verify the entity identifier group determined by the entity identifier group determiner.

The triple storage module stores entity identifier group summary information and edited entity identifier group summary information for each entity.

 The second grouping unit loads the same attribute mapping information and the representative attribute for each entity type from the previously stored attribute information table, and regroups the groups according to the entity type based on the representative attribute of each entity identifier.

In addition, the second grouping unit obtains the similarity of the character string corresponding to the attribute value based on the representative attribute name of the entity identifier for each entity type, and groups the entity identifiers having the determined similarity having a predetermined threshold value or more into the same group.

The entity identifier group determination unit obtains the association attribute values of the entities in each group from the multiple ontology database, checks whether there is an entity identifier having the same attribute value among the association attribute values of all entities, and identifies an entity identifier having the same attribute value. If present, group object identifiers with the same attribute value into subgroups.

According to another aspect of the present invention, a method for identifying an individual by an object identification device, comprising: (a) acquiring attribute information of the new individual identifier from a multiple ontology database when a new individual identifier is input, (b) Grouping the new entity identifier based on an entity type using an attribute representing a type among the acquired attribute information, (c) the entity identifier based on representative attributes of entity identifiers included in a group according to each entity type (D) acquiring and comparing the association attribute values for the entities of each of the regrouped groups, and dividing each group into subgroups to determine an entity identifier group based on a result of the comparison. An object identification method using multiple ontology including a is provided.

The object identification method using the multiple ontology repeats the step of determining the entity identifier group for the new entity identifier after step (d) to perform verification of the entity identifier group determined by the entity identifier group determiner. It may further include.

In the step (c), the same attribute mapping information and the representative attribute are loaded for each entity type from the prestored attribute information table, and the groups according to the entity type are grouped again based on the representative attribute of each entity identifier.

Also, in the step (c), the similarity of the string corresponding to the attribute value is obtained based on the representative attribute name of the entity identifier for each entity type, and the entity identifiers having the determined similarity having a predetermined threshold value or more are grouped into the same group.

In the step (d), the association attribute values of the entities in each group are obtained from the multiple ontology database, and whether the entity identifiers having the same attribute values are found among the association attribute values of all the entities. If present, group object identifiers with the same attribute value into subgroups.

As described above, according to the present invention, various types of entities may be gradually identified by using attribute information of multiple ontologies included in linked data.

In addition, the ontology is grouped into the same type and object by using the property information of multiple ontology, and the object is identified through the comparison of the main property of each group to improve the accuracy of identifying the same object and added by comparing with the external service. Only URIs can be supported for progressive object identification.

1 is a view showing an individual identification system using multiple ontologies according to the present invention.

Figure 2 is a block diagram schematically showing the configuration of the object identification apparatus according to the present invention.

3 is a flowchart illustrating a method for identifying an object by the object identifying apparatus according to the present invention.

Details of the above-described objects and technical configurations of the present invention and the effects thereof according to the present invention will be more clearly understood by the following detailed description based on the accompanying drawings.

First, the terms used in the present invention will be described.

Linked Data is a movement and code of conduct for the data network at the heart of the next generation of the Web. The goal is to open and connect data freely on the web, and to enable these data to collaborate again to realize a true data web. Therefore, HTTP (hypertext transfer protocol) is used for data distribution on the web, and Resource Description Framework (RDF) and Simple Protocol and RDF Query Language (SPARQL) are used to ensure connectivity and accessibility, which is included in the ontology. This is a standard query language used for querying collected information. It performs functions similar to the structured query language (SQL) of the database management system (DBMS).

In addition, Linked Data names specific concepts as Uniform Resource Identifiers (URIs), allows access to resources named as URIs over HTTP, and provides detailed information contained by RDF-based URIs when accessed. It provides access to other relevant concepts contained in RDF.

URIs are identifiers of classes, properties, and entities that are mapped to unique addresses on the Web to share information resources over the Web. In other words, it is an internet address and identifier used to access information of a class, property, or object through a URL. The URI is used in the term entity identifier.

Ontology is constructed by defining concepts and attributes and establishing the relationship between concepts and concepts by obtaining knowledge about the components of ontology that is to be constructed in ontology from related documents of specific field, namely concept, property of concept, connection relation between concepts. In the semantic web, concepts are represented by URIs. For example, the "Person" class is represented in the ontology as a URI such as "http://www.etri.re.kr/example#Person".

The "Person" class has properties such as name, age, birthplace, and so on. The "name" attribute can have a string as its value, the "age" attribute can have an integer as its value, and the "birthplace" attribute can have an instance of "Location" as its value. have.

Entity identification refers to the explicit establishment of a relationship to a URI in order to ensure that the classes, properties, and entities that make up the ontology are interrelated. For example, if the A-1 entity of ontology A is the same as the B-1 entity of ontology B, you can merge A-1 and B-1 related properties by specifying that the A-1 and B-1 entities have the same relationship. To help.

1 is a view showing an individual identification system using multiple ontologies according to the present invention.

Referring to FIG. 1, an object identification system using multiple ontologies includes multiple ontology databases (300a, ..., 300n, hereinafter 300) in which attribute information of each individual is stored as an ontology, and an entity associated with a specific URI (URI). It includes the object identification device 100 for performing the object identification to set the relationship to the.

The multiple ontology database 300 may be a server that provides services such as Sindice.com service, sameAs.org service, and the like. Here, the Sindice.com service is a semantic web search engine, which continuously collects various types of ontology in real time and, when an object name is input, provides a URI and an associated ontology with the corresponding object name as a search result.

The sameAs.org service refers to a service that collects URIs representing the same entity relationships collected from various ontologies and provides a group of previously identified results when searching for entity names or URIs.

The object identifying apparatus 100 identifies the objects by obtaining and comparing attribute information of the object to be identified from the multiple ontology database when a new object identifier requiring object identification is input.

That is, the object identifier device 100 obtains and compares the object identifiers by obtaining and comparing the attribute information of the object identifiers included in the object identifier list from the multiple ontology including the attribute information of the object with respect to the object identifier list to be identified. And generate an entity identifier group as a final result of the entity identification.

In addition, the object identification device 100 gradually identifies various types of entities by using attribute information from the multiple ontology database 300 included in the linked data.

In addition, the object identification device 100 collects the attribute information of the new entity identifier from the multiple ontology database 300 through the SPARQL endpoint, the attribute information of the new entity identifier and the identification result obtained from the triple storage module. Perform entity identification using. Here, the SPARQL endpoint refers to a service point of contact that allows access to the multiple ontology database 300, and provides the ontology information in RDF or various other formats through a web service.

Detailed description of the object identification device 100 performing the above role will be described with reference to FIG. 2.

2 is a block diagram schematically showing the configuration of the object identification apparatus according to the present invention.

Referring to FIG. 2, the object identification apparatus 100 may include a communication unit 110 for communicating with a multiple ontology database, a user interface unit 120 for receiving a query from a user, a triple storage unit 130, and attribute information collection unit. 151, a first grouping unit 152, a second grouping unit 153, an object identifier group determining unit 154, and a verification unit 155.

The triple storage unit 130 stores object identification results for each object in triple form. That is, the triple storage unit 130 stores entity identifier group summary information and edited entity identifier group summary information for each entity. The entity identifier group summary information includes the entity number, entity type, representative entity name, representative entity identifier, etc. for each entity identifier group.

In addition, the triple storage unit 130 stores information on the object type, representative attribute values for each type, and lower attributes to be considered together in an ontology form. That is, the triple storage unit 130 stores only the information and the identification result necessary for identifying the individual among the multiple ontology information, and other information is obtained by accessing the multiple ontology database in real time.

The attribute information collecting unit 151 collects attribute information of a new entity identifier from the multiple ontology database. For example, when the new entity identifier is a URI for "Hong Gil-dong", the attribution information collecting unit 151 collects attribution information such as age, occupation, place of birth, e-mail address, etc. for Hong Gil-dong from the multiple ontology database. do.

The first grouping unit 152 groups the new entity identifier according to the entity type by using the attribute representing the type among the attribute information collected by the attribute information collecting unit 151. That is, the first grouping unit 152 determines the entity type of the new entity identifier by using an attribute representing a type among the collected attribute information, and determines the entity type of the entity identifier group from the entity identifier group for the query. A corresponding group is selected and assigned to the group of the new entity identifier.

The second grouping unit 153 loads the same property mapping information and the representative property for each group according to the object type from the previously stored property information table, and regroups the groups according to the object type based on the representative property of each object. do. That is, the second grouping unit 153 calculates the similarity of the string corresponding to the attribute value based on the representative attribute name of the entity identifier for each object type, and sets the entity identifiers having the predetermined similarity or more to a same group as the same group. Group. For example, in the case of a person, since the representative attribute is a human name, the similarity of the string corresponding to the human name is calculated, and the individual identifiers whose calculated similarity has a predetermined threshold or more are made into the same group.

The entity identifier group determiner 154 obtains and compares an associated attribute value for each group of entities grouped by the second grouper 153 from the multiple ontology database, and based on the comparison result, Divide the group into subgroups to determine the group of individual identifiers. That is, the entity identifier group determiner 154 obtains the association attribute values of the entities in each group from the multiple ontology database, checks whether there is an entity identifier having the same attribute value among the association attribute values of all entities, If there is an object identifier with a value, group the object identifiers with the same attribute value into a subgroup. In this case, when there is no entity identifier having the same attribute value, the second grouping unit 153 maintains the grouped group.

For example, in the case of an entity identifier (URI) having a representative attribute value of "Hong Gil-dong", the entity identifier group determination unit 154 may determine "work", "association attribute values of" Hong Gil-dong "from the multiple ontology database. e-mail address "is obtained, and Hong Gil-dong having the same" attribute "and" e-mail address "as the related attribute values is made into the same group. By the above method, the entity identifier group determiner 154 may group entity identifiers having the same attribute value into subgroups.

The verification unit 155 repeats the process of determining the entity identifier group for the new entity identifier and verifies the entity identifier group determined by the entity identifier group determiner 154.

3 is a flowchart illustrating a method for identifying an object by the object identifying apparatus according to the present invention.

Referring to FIG. 3, the apparatus for identifying an object obtains attribute information of a new object identifier to be identified from a multiple ontology database (S302), and uses the attribute representing a type in the obtained attribute information according to the object type. The group is assigned to (S304). That is, the object identifying apparatus checks an attribute capable of distinguishing the type of the object through an upper and lower relationship with the rdf: type or the explicit type class among the attribute information of the new object identifier, and checks the new object type as the new object identifier. Assign an object identifier.

For example, referring to Table 1, if a subclass of foaf: person exists, the subclass can be determined to be a class meaning a lifespan. That is, if kisti: person, class 1 belonging to Ontology A, is a subclass of foaf: person, the object identifier can be used to determine the type of the object later by identifying and storing kisti: person as the object type corresponding to the person. do.

Table 1 kisti: person rdf: type Identity: Person foaf: person rdf: type Identity: Person kisti: institution rdf: type Identity: Institution

When S304 is performed, the entity identification device loads the same attribute mapping information and the representative attribute for each group grouped according to the entity type in the attribute information table (S306). That is, the object identification device may use the same attribute mapping information (eg, kisti: hasCreator and foaf: maker are identical attributes) and representative attributes (for example, between attributes of multiple ontologies to be used for identification according to the type of object to be identified). For example, the representative attribute of a person loads the “foaf: name” attribute in the foaf ontology.

When the step S306 is performed, the entity identification device regroups the groups grouped by the entity type based on the representative attribute (S308). That is, the entity identification device calculates the similarity of the string corresponding to the attribute value based on the representative attribute name of the entity identifier for each entity type, and groups the entity identifiers having the predetermined similarity greater than or equal to a predetermined threshold value into the same group.

For example, in the case of people, the similarity is calculated for the human name and in the case of the company, and grouped using the calculated result. In this case, we use the Jaro Winkler Distance string comparison method, which is suitable for relatively short string comparisons.

 In addition, the object identification device forms a representative attribute name in a predetermined format according to a predetermined rule and then groups the similar attribute by calculating similarity. For example, in the case of human names, the similarity is calculated through string comparison considering the First Name, Middle Name, and Last Name.

In addition, in the case of the institution name, similarity is calculated through string comparisons considering notation names such as ".inc", "co", and "INC".

After performing S308, the entity identification device obtains and compares the associated attribute values of the entity identifiers belonging to each group from the multiple ontology database, and divides each group into subgroups based on the comparison result to divide the entity identifier group. Determine (S310).

That is, the entity identification device obtains an association attribute value for an entity identifier belonging to each group from the multiple ontology database and checks whether entity identifiers having the same association attribute value exist. When there are entity identifiers having the same attribute value as the result of the checking, the entity identifier device determines the entity identifier group by dividing the entity identifiers having the same attribute value into subgroups.

Through this process, the entity identification device identifies mappings between entity identifiers classified into different groups or entity identifiers having the same group but no mapping property in the process of grouping the entity identifiers based on the previous entity type or entity name.

That is, when the representative attributes are mapped to the same association attribute value A of the entity identifiers 1 and 2 belonging to different groups, the entity identification device obtains the association attribute value of A, and the entity identifiers 1 and 2 are all associated attributes of A. Check if it is included.

For example, if the object identifiers 1 and 2 are different author names, but all have the author relationship of the A paper, the object identification device collects the association attributes of the A paper, and the individual identifiers 1 and 2 are all linked in the author relationship. Check if it is.

When the S310 is performed, the object identifying apparatus verifies the object identification result by repeating the steps S502 to S510, and finally confirms the object identifier group (S312).

As such, those skilled in the art will appreciate that the present invention can be implemented in other specific forms without changing the technical spirit or essential features thereof. Therefore, the above-described embodiments are to be understood as illustrative in all respects and not as restrictive. The scope of the present invention is shown by the following claims rather than the detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalents should be construed as being included in the scope of the present invention. do.

As described above, the apparatus and method for identifying individuals using multiple ontology of the present invention is suitable for the high necessity of gradually identifying various types of entities using attribute information of multiple ontology included in Linked Data.

Claims (11)

A triple storage unit storing entity identification results for each entity in triple form; An attribute information collection unit for obtaining attribute information of the new entity identifier from a multiple ontology database when a new entity identifier is input; A first grouping unit for grouping the new entity identifier based on an entity type using an attribute representing a type among the acquired attribute information; A second grouping unit for regrouping the entity identifiers based on representative attributes of the entity identifiers included in the group according to each entity type; And An entity identifier group determination unit which obtains and compares an association attribute value for entities of each group grouped by the second grouping unit, and determines an entity identifier group by dividing each group into subgroups based on the comparison result; Individual identification device using multiple ontology, characterized in that it comprises a. The method of claim 1, And a verification unit configured to repeat the process of determining the entity identifier group for the new entity identifier and to verify the entity identifier group determined by the entity identifier group determiner. The method of claim 1, The triple storage module is an object identification device using multiple ontology, characterized in that the object identifier group summary information, the edited object identifier group summary information for each object is stored. The method of claim 1, The second grouping unit loads the same property mapping information and the representative property for each object type from a pre-stored property information table, and re-groups groups according to the object type based on the representative property of each object identifier. Object identification device using. The method of claim 4, wherein The second grouping unit obtains the similarity of the string corresponding to the attribute value based on the representative attribute name of the entity identifier for each entity type, and groups the identifiers having the determined similarity with a predetermined threshold value or more into the same group. Individual identification device using ontology. The method of claim 1, The entity identifier group determiner, Obtain the related property value of the objects in each group from the multiple ontology database, check whether there is an object identifier having the same property value among the related property values of all the objects, and if there is an object identifier having the same property value, the same property value Individual ontology identification apparatus using multiple ontology, characterized in that to bind the object identifiers having a subgroup. In the method of identifying the object, (a) when the new entity identifier is input, obtaining attribute information of the new entity identifier from a multiple ontology database; (b) grouping the new entity identifier based on the entity type using an attribute representing a type among the acquired attribute information; (c) regrouping the entity identifiers based on representative attributes of the entity identifiers included in the group according to each entity type; And (d) acquiring and comparing the association attribute values for the entities of each of the regrouped groups, and dividing the groups into subgroups to determine an entity identifier group based on the comparison result; Individual identification method using multiple ontologies comprising a. The method of claim 7, wherein After the step (d), Repeating the determination of the entity identifier group for the new entity identifier further comprises the step of performing the verification of the entity identifier group determined by the entity identifier group determiner further comprising the individual ontology. The method of claim 7, wherein In step (c), Object identification using multiple ontology, characterized in that to load the same attribute mapping information and the representative attribute by the object type from the stored attribute information table, and to group the groups according to the object type based on the representative attribute of each object identifier. Way. The method of claim 9, In the step (c), the similarity of the string corresponding to the attribute value is obtained based on the representative attribute name of the entity identifier for each entity type, and the entity identifiers having the determined similarity having a predetermined threshold value or more are grouped into the same group. Individual identification method using multiple ontology. The method of claim 7, wherein In step (d), Obtain the related property value of the objects in each group from the multiple ontology database, check whether there is an object identifier having the same property value among the related property values of all the objects, and if there is an object identifier having the same property value, the same property value Individual identification method using multiple ontology, characterized in that to bind the object identifiers having a subgroup.

Images