메시지 전달 기법을 이용한 개념도 구축 시스템의 설계 및 구현

Design and Implementation of a system for constructing concept maps by passing messages between concepts

초록

Concept map is a graphical tool that is widely used for organizing and representing knowledge and shows the relationships among related concepts. Automatic concept map construction from text documents requires methods for extracting concepts and relationships (taxonomic and non-taxonomic). Even though a lot of studies have been conducted to automatically construct a concept map, they still have some limitations such as a resolution of anaphora problem and defining relationships to form propositions. In this paper, we propose a clustering-based approach for constructing a concept map from text documents. First, relevant concepts are extracted using typed dependency linguistic patterns. Anaphoric resolution for pronouns is then introduced to map the pronouns with candidate terms. Second, extracted concepts are clustered using affinity propagation algorithm. Finally, relationships are assigned between the extracted concepts in each cluster. Our empirical results show that the constructed concept maps conform to the outputs generated manually by domain experts. Furthermore, domain experts verified that the constructed concept maps are in accordance with their knowledge.

키워드

개념도; 개념도 학습; 지식 획득; 친근도 전파; 텍스트 클러스터링; Concept Map; Concept Map Learning; Knowledge Acquisition; Affinity Propagation
제목
메시지 전달 기법을 이용한 개념도 구축 시스템의 설계 및 구현
제목 (타언어)
Design and Implementation of a system for constructing concept maps by passing messages between concepts
저자
이크발 카심; 이동호; 정진우; 허지욱
발행일
2013-04
저널명
데이타베이스연구
권
29
호
1
페이지
51 ~ 71