A learning method of Chinese embodiment of non-taxonomy relations in instances

作者：Liu Jie^*; Peng Yan; Wang Suozhu; Liu Lizhen; Huang Xiangyang; Wang Xuren

来源：Journal of Computational Information Systems, 2010, 6(1): 167-173.

• Chinese sentence
• Corresponding weights
• In-process
• Information Extraction
• Learning methods
• Non-taxonomy relations
• Ontology learning

Learning of ENTRI(embodiment of non-taxonomy relations in instances)is an essential part of ontology learning, but at present research on it has rarely been carried out. Given that learning of ENTRI has the same purpose of the extraction of entities relations as information extraction and Chinese sentences have flexible structure and richly ideographic words, the snowball model is improved to extract ENTRI. In process of extracting ENTRI, Hownet is used to compute the similarity between words, and different parts of speech is given corresponding weight according to its importance during computation of similarity between sentences, in addition the computational method of similarity between sentences is improved because Chinese sentences have flexible structure. The experiment results show the performance of the proposed method is quite good and can play a good assistant role in extraction of ENTRI.

• 出版日期2010
• 单位首都师范大学