Domain ontologies contain information about the important concepts in a domain, the associated attributes and the relationships between various concepts. The manual creation of domain ontologies is an expensive and time consuming process. In this paper, we present an approach to the automatic extraction of domain ontologies from domain-specific text. This approach uses dependency relations between terms to represent text as a graph. The graph based ranking algorithm HITS is used to identify domain keywords and to structure them as concepts and attributes. Experiments on two different domains, digital cameras and wines, show that our method performs well in comparison to other approaches.
展开▼
