Preliminary Chinese Term Classification for Ontology Construction
Gaoying Cui, Qin Lu, Wenjie Li · PolyU Institutional Research Archive (Hong Kong Polytechnic University) · 2008
An ontology can be seen as a representation of concepts in a specific domain. Accordingly, ontology construction can be regarded as the process of organizing these concepts. If the terms which are used to label the concepts are classified before building an ontology, the work of ontology construction can proceed much more easily. Part-of-speech (PoS) tags usually carry some linguistic information of terms, so PoS tagging can be seen as a kind of preliminary classification to help constructing concept nodes in ontology because features or attributes related to concepts of different PoS types may be different. This paper presents a simple approach to tag domain terms for the convenience of ontology construction, referred to as Term PoS (TPoS) Tagging. The proposed approach makes use of segmentation and tagging results from a general PoS tagging software to predict tags for extracted domain specific terms. This approach needs no training and no context information. The experimental results show that the proposed approach achieves a precision of 95.41% for extracted terms and can be easily applied to different domains. Comparing with some existing approaches, our approach shows that for some specific tasks, simple method can obtain very good performance and is thus a better choice.