CSDL Home W WI-IAT 2011 2011 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies
Aug. 22, 2011 to Aug. 27, 2011
Many NLP and IR applications require semantic classification knowledge of words. However, manually constructing semantic classes is a time-consuming and labor-intensive task. In this paper, we present an algorithm for induction of Chinese semantic classes from natural language text based on coordinate patterns. First, several coordinate patterns are proposed to harvest high-quality coordinate instance. Second, an iterative clustering process is used to cluster words into semantic classes. The clustering process mainly used coordinate relation between words. Experiment results show that the proposed approach performs relatively well and achieves 53.2% in terms of precision. Finally, a thesaurus containing about 15000 Chinese words is generated automatically.
semantic class, coordinate structure, bottom-up clustering, language resource
Likun Qiu, Yunfang Wu, Jing Shi, Yanqiu Shao, Zhiyi Long, "Induction of Semantic Classes Based on Coordinate Patterns", WI-IAT, 2011, 2011 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies, 2011 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies 2011, pp. 201-204, doi:10.1109/WI-IAT.2011.66