Ontology Construction And Its Applications In Question-Answering System

As a semantic foundation of concepts in a domain, ontology has been widely applied to information retrieval, artificial intelligence, knowledge network, knowledge management etc. However, ontology construction is a time-consuming and tedious task, so automatic or semi-automatic methods have been proposed by some researchers. In this paper, the current research states and progresses of automatic ontology construction technologies are summarized and some FCA based methods are deeply analyzed.The model proposed in this paper is mainly composed of three steps: (1) PAT-Tree based phrase extraction; Compared with other segmentation methods, statistic method is capable of identifying phrases which doesn't exit in dictionary. (2) Feature selection; High dimension of text is always a great problem for applications like text clustering and classification, so feature selection is necessary. (3) Formal context acquirement and FCA based ontology construction. Formal context is the binary relation between text and its features. FCA builds its concept lattice from this context, which will produce ontology for text classification.Experiments results show that FCA based method can improve automatic construction of ontology and the constructed ontology by this model is readable and reliable. Finally, the ontology constructed by this method is tested on question-answering system. The result proved that this semi-automatic method is effective in supporting question-answering system and automatic degree of ontology construction. But human effort is needed to facilitate this method in order to acquire higher precision. Compared with manual ontology, this method reveals an equally satisfying result. And it is worthwhile for automatic ontology construction research.