KR20130045425A

KR20130045425A - Expert recommendation search method based on social ontology

Info

Publication number: KR20130045425A
Application number: KR1020110109097A
Authority: KR
Inventors: 양재동
Original assignee: (주)케이테크
Priority date: 2011-10-25
Filing date: 2011-10-25
Publication date: 2013-05-06

Abstract

The present invention relates to a social ontology-based knowledge expert recommendation method that can easily recommend a highly qualified group of experts for acquiring and verifying expertise required for production of knowledge from an expert social ontology, which includes: a) receiving mentoring on expertise. Receiving, via a social network, a natural language search query relating to an expert recommendation; b) extracting and determining knowledge-based keywords by analyzing the natural language search query; c) determining a target specialization related to the knowledge-based keyword through inference based on a social ontology layer composed of a specialty social ontology layer and an expert social ontology layer assigned to an expert for each specialty of the specialty social ontology. Steps; and d) extracting an expert assigned to the target specialty from the expert social ontology layer among the social ontology.

Description

Expert recommendation search method based on social ontology

The present invention relates to a social ontology-based knowledge expert recommendation method that can be easily recommended from the expert social ontologies of the highly-satisfied expert group for obtaining and verifying the expertise required for knowledge production.

Ontology is defined in computer science and information science as a data model representing a specific domain. It is defined as a set of formal vocabulary describing the concept belonging to a specific domain and the relationship between the concepts. For example, if you describe in formal terms vocabulary, the relationship between species, and the relationship between English words, which are classified as "species, genus, thu, thu, gang-mun-system", each can be called an ontology. Ontology, a set of vocabulary described in formal language, is used for reasoning. The emergence of the web has brought about a revolution in all aspects of society, including traditional information retrieval, knowledge management and general commerce. In particular, web information retrieval has made it possible to search electronic resources that can be accessed through the web in a limited search of the collections. Due to the rapid development of the web, the expansion of search targets required more sophisticated searches, which prompted the development of intelligent information retrieval systems. Based on these momentum, the necessity of a new tool for information retrieval to effectively manage web resources has emerged. Ontology is a tool that can implement semantic web, and it is a tool that can connect knowledge concept semantically.

Ontology is used in the field of machine translation and artificial intelligence of natural language. Recently, it is attracting attention as a key element such as the semantic web and the semantic web service derived from it. have.

The definition of ontology, often cited, is 'An ontology is an explicit and formal specification of a conceptualisation of a domain of interest.' (Cf Tom Gruber, 1993). In other words, in order to find common ground in each thing and to express it as a set or category, it is to describe clearly and in detail the meaning, use of knowledge, etc. Borst is a formal specification of 'shared concepts.' Ontologies are defined as a formal specification of a shared conceptualization.) The concept of 'shared' has been added to the previous definition, which means that a well-defined concept is commonly used in each field. Can be.

Such ontology is widely applied to the field of artificial intelligence, information retrieval, ubiquitous computing, and electronic commerce. In particular, the ontology of the information retrieval field, which is one of the representative fields, can prevent unnecessary errors and improve retrieval efficiency only by using a term collection or a synonym dictionary. For example, a user's mismatched keyword 'equality' will be corrected to 'unfair' using ontologies, and richer search services will be available using similar or related terms such as 'unfair competition, monopoly, dumping, and government subsidies.' It can be provided. In the open directory project, voluntary participants make a classification system of Internet information, which is used as a representative web information classification system that can be used in numerous sites including commercial search sites such as Google. Increasingly, classification categories are available.

On the other hand, a social network ontology system that provides an ontology inference has been proposed in the Patent Publication No. 10-2011-0058982. Although the expert / mentor reasoning section is mentioned in this system, it is stated that the ontology can be used to search the experts and surrounding data related to the experts. There is no description of a method for determining a keyword indicating a field using an ontology.

The present invention provides a social ontology-based knowledge expert recommendation method that can be easily recommended from the expert social ontology even if the expert group for each field with high satisfaction for acquiring and verifying the expertise required for producing the knowledge does not exactly know the field to which the expert belongs. There is this.

The present invention for achieving the above object,

a) receiving, via a social network, a natural language search query for expert recommendation that can be mentored to the expertise;

b) extracting and determining knowledge-based keywords by analyzing the natural language search query;

c) determining a target specialization related to the knowledge-based keyword through inference based on a social ontology layer composed of a specialty social ontology layer and an expert social ontology layer assigned to an expert for each specialty of the specialty social ontology. Steps;

d) extracting an expert assigned to the target specialty from the expert social ontology layer among the social ontology; and providing a social ontology-based knowledge expert recommendation method.

In particular, in step b), when two or more knowledge-based keywords are extracted from the search query, the step of extracting a connection word connecting a relationship between adjacent knowledge-based keywords, and step c) reflects the connection word, reflecting the target specialized field. It is good to decide.

In the step c), if the connection word extracted in step b) corresponds to a connection word having the meaning of 'and', the common or related knowledge terms among the lower level knowledge terms of the knowledge terms related to each knowledge-based keyword are used. As the target specialization,

When the connection word corresponds to a connection word having the meaning of 'or', it is desirable to determine a knowledge term that is common or related among the higher level knowledge terms of the knowledge word related to each knowledge-based keyword as the target specialization.

In addition, the step of determining the ranking of the expert according to the degree of expertise associated with the target specialization field for the expert extracted from the expert social ontology layer, in particular, the degree of expertise is the number of publications, research reports and patents It is preferable to evaluate on the basis of that.

Social ontology-based knowledge expert recommendation method of the present invention has an effect that can be easily recommended from the expert social ontology, a group of experts with high satisfaction to obtain and verify the expertise required for knowledge production.

1 is a flowchart schematically illustrating a social ontology-based knowledge expert recommendation method.
FIG. 2 is a diagram schematically illustrating a social ontology consisting of a specialized social ontology layer and an expert social ontology layer.
3 is a diagram illustrating a natural language search query on a social network.
4 and 5 are diagrams illustrating a state in which a target specialization is determined by a natural language search query.
5 is a diagram illustrating a state where an expert is recommended according to a search query.

Hereinafter, the social ontology-based knowledge expert recommendation method of the present invention will be described in detail with reference to Examples. The scope of the present invention is not limited to the following Examples.

The social ontology-based knowledge expert recommendation method of the present invention includes a natural language search query reception step, a knowledge-based keyword extraction step, a target specialization field determination step, and an expert search step.

1 is a flowchart schematically illustrating a social ontology-based knowledge expert recommendation method.

The receiving of the search query is a step of receiving a natural language search query related to a recommendation of an expert who can receive mentoring on expertise through a social network.

For example, if a knowledge consumer makes a search query 'Recommend experts on web application platforms using Eclipse' through a social network, the search query is received by the management server through the network.

Next, the search query received by the management server is analyzed to extract and determine knowledge-based keywords.

The search query analyzer of the management server extracts knowledge-based keywords from the search query. In this case, when two or more knowledge-based keywords are extracted by the search query analyzer, a connection word for connecting a relationship between adjacent knowledge-based keywords is extracted.

In the case of the above search query, the knowledge base keywords are 'Eclipse' and 'Web application platform experts', and the connection word is 'use'.

In this case, when analyzing the knowledge-based keyword, the search query analyzer reflects all the synonyms and the like.

The target specialty related to the knowledge-based keyword is determined through the inference unit of the social ontology built in the management server. When determining the target specialization, not only knowledge-based keywords but also synonyms thereof are reflected.

As shown in FIG. 2, the social ontology 1 includes a specialty social ontology layer 1a and an expert social ontology layer 1b to which an expert is assigned for each specialty of the specialty social ontology.

When two or more knowledge-based keywords are extracted, the target specialization is determined by reflecting a connection word connecting the relationship between adjacent knowledge-based keywords.

When the link word corresponds to a link word having a meaning of 'and' such as 'high', 'high', 'and', 'and', 'using', 'using', etc. The knowledge of common or related knowledge terms among the lower level knowledge terms is determined as the target specialization.

For example, if the search query is' recommended expert knowledge about dynamic web and RIA 'as shown in Fig. 3, the search query analyzer is a knowledge-based keyword such as' dynamic web (dynamic web platform)' and 'RIA (RIA platform). ), And 'and' as the connecting word.

In addition, since the connection word corresponds to a connection word having a meaning of 'and', common 'siliver light' and 'rich' among the lower level knowledge terms 10 of 'dynamic web platform' and 'RIA platform' as shown in FIG. Client 'is determined as the final target specialization 110.

And if the connection word corresponds to a connection word having the meaning of 'or' such as 'or', 'me', 'or', 'or', etc., among the higher level knowledge terms of the knowledge term related to each knowledge-based keyword Determine relevant or relevant knowledge terms as target specializations.

For example, if the search query is "Recommend experienced users who have used Xerces as an expert in the dynamic web or RIA?", The search query analyzer is a knowledge-based keyword called 'Dynamic Web (Dynamic Web Platform)', 'RIA'. (RIA platform) ','Xerces' and 'or' and 'use' as linking words. As shown in FIG. 5, since the first connection word corresponds to a connection word having the meaning of 'or', 'web application platform' which is a high level knowledge term of 'dynamic web platform' and 'RIA platform' (30) ). Next, because the second term 'use' is related to Xerces, the term 'use' is associated with Xerces or the higher level of DOM, which is the terminology included in the domain of the web application platform (310). Is determined as the final target specialization 330.

Next, as shown in FIG. 5, the expert 50 assigned to the target specialty is extracted from the expert social ontology layer 1b of the social ontology 1. The extracted expert is displayed on the web as shown in FIG. 6, where the specialty ontology layer of the social ontology is displayed together.

In this case, it is desirable to display the experts along with the ranking according to the degree of expertise related to the target specialization. The degree of expertise may be evaluated based on the number of articles, article citations, research reports, and patent publications.

1: social ontology,
1a: specialty social ontology layer,
1b: Professional Social Ontology Layer,
10: lower level,
110, 330: target specialization,
30: higher level,
310: related links

Claims

a) receiving, via a social network, a natural language search query for expert recommendation that can be mentored to the expertise;
b) extracting and determining knowledge-based keywords by analyzing the natural language search query;
c) determining a target specialization related to the knowledge-based keyword through inference based on a social ontology layer composed of a specialty social ontology layer and an expert social ontology layer assigned to an expert for each specialty of the specialty social ontology. Steps;
and d) extracting an expert assigned to the target specialty from the expert social ontology layer among the social ontology.

The method of claim 1,
In step b), if two or more knowledge-based keywords are extracted from the search query, extracting a connection word connecting the relation between adjacent knowledge-based keywords,
Step c) is a social ontology-based knowledge expert recommendation method, characterized in that for determining the target specialization by reflecting the connection word.

The method of claim 2,
In step c), if the connection word extracted in step b) corresponds to a connection word having the meaning of 'and', the common or related knowledge terms among knowledge terms of the lower level knowledge terms related to each knowledge-based keyword are used. Social ontology-based knowledge expert recommendation method, characterized in that the step of determining the target specialization.

The method of claim 3,
In step c), if the connection word extracted in step b) corresponds to a connection word having an meaning of 'or', the common or related knowledge terms among the higher level knowledge terms of knowledge terms related to each knowledge-based keyword are used. Social ontology-based knowledge expert recommendation method, characterized in that the step of determining the target specialization.

The method of claim 1,
Social ontology-based knowledge expert recommendation method comprising the step of determining the ranking of the expert in accordance with the degree of expertise associated with the target specialization field for the expert extracted from the expert social ontology layer.

The method of claim 5,
The degree of expertise is a social ontology-based knowledge expert recommendation method characterized in that the evaluation based on the number of publications, research reports and patents.