Automatic Extraction Of Conceptual Relations For Constructing Domain-Specific Ontology

Ontology is a shared knowledge representation to describe concepts and their relations. It has attracted lots of researchers to focus on this field since it was proposed. And it has been widely applied to many areas of computer science. Although several ontology development tools are available, huge efforts are required to collect concepts and their relations, which make it hard for ontologies to apply for real uses. To apply machine learning techniques for ontology construction will largely speed up the process, in hence save much time and human resources. So ontology learning has become a hot research topic.Our research processes Chinese free text, and combines statistic and rule-based methods to learn relations from closed corpus and open corpus in a semi-supervised way. The experiments indicate this approach is valid and efficient, and portable to other domains.