Ontology Learning for Chinese Information Organization and Knowledge Discovery in Ethnology and Anthropology

 

Kong Jing

 

Institute of Ethnology & Anthropology, Chinese Academy of Social Sciences,

Beijing, 100081, China

 

Abstract:

This paper presents an ontology learning architecture which reflects the interaction between ontology learning and other application such as ontology-engineering tools and information system. It extracts knowledge from information as well as performs semantic annotation of information so that forms a cycle process of interacting between information flow and knowledge flow. Based this architecture, we developed a prototype system CHOL: an ontology learning system for Chinese Information Organization and Knowledge Discovery. CHOL learns domain ontology from Chinese scholarly literature. It supports to semi-automatic domain ontology acquisition and dynamic maintenance, on the other hand, supports to auto-indexing and auto-classification of Chinese scholarly literature. CHOL has been applied in Ethnology and Anthropology for Chinese Information Organization and Knowledge Discovery.

In CHOL, we built a- Chinese domain ontology which includes 5 sub-ontologies: Natural Language Ontology, Global Domain Ontology, Foundational Domain Ontology, Specific Domain Ontology and Domain Ontology Instances. This domain ontology focused on the reuse, function, completeness and consistency. Here, we used Hownet to build Natural Language Ontology and used Chinese Classification Thesaurus to build Global Domain Ontology. We proposed a pursuing high precision multi-strategy method of extracting Chinese domain concepts and a hybrid several similarity measures method of domain term identifying which utilizes Global Domain Ontology to compute the similarity measures between a term and a specific domain. We also proposed a method for acquisition of concepts relationship based neural networks in CHOL. This method can learn multiple concepts of a term in different subjects and concept relationships for each concept. These methods we proposed had been proved to be feasible and effective.

 

Keyword: ontology learning         domain ontology   automatic ontology acquisition                         concepts extraction

 

The Author:  Kong Jing, Female, 1969. Research interests include information system, information retrieval and knowledge discovery in data and text. E_mail:kongjing@cass.org.cn