Research On Automatic Construction Of Chinese Domain Ontologies

Ontologies have shown their usefulness in application areas such as intelligent information integration, information brokering and natural-language processing, to name but a few. However, their wide-spread usage is still hindered by ontology engineering being rather time-consuming and, hence, expensive. Automatic ontology construction has become the key point in this field.In this paper, we present a method that automatically mine an ontology from a large corpus in a specific domain. The research work and innovation are as follows.1) Research on ontology constructing methodologiesSeveral known methods of ontology construction and their features are analyzed respectively. The key technologies for automatic ontology construction are researched, including term extracting, concept learning and relation mining.2) Automatic term extractionWe propose a novel method that is based the probability contrast of the term appearance in domain corpus and background (balance) corpus. The log-likelihood ratio (LLR) is used to score terms, which is effective.3) Concept learning and relation miningThe G-N algorithm, which is based on the "Small World Feature" in the Complex Network, is successfully introduced to the field of Statistic Natrual Language Processing. The G-N algorithm is used to achieve the cluster analyzing of the term network to mining concepts and relations. Hownet is adopted as a supplement to the terms' similarities during building the term network.4) Automatic domain ontology construction systemWe have implemented the prototype system of automatic domain ontology construction by the key technologies. In experiment an ontology of computer science domain has been constructed.