A new method for automatically constructing domain-oriented term taxonomy based on weighted word co-occurrence analysis

The automatically construction of term taxonomy can enhance our ability for expressing the science mapping. In this paper, we introduce the definition of weighted co-occurring word pair and corresponding improved method of word co-occurrence analysis. An application and evaluation of this proposed method in the library and information science is also discussed, which includes how to get the expanded effective keywords, how to calculate the weight of keywords and their relations, and how to abstract the hierarchical structures and other relations such as synonyms and etc. A visualization tool and a prototype search system are designed for browsing the term taxonomy identified. Finally, we report the experiment of evaluation and comparison. The experiment results prove that this proposed method in helping users doing semantic searches and expanding their searching results is effective and can meet the requirement of some specific domains.