基于支持向量机的日语并列关系解析

日语并列关系是日语依存关系的一种.传统的日语依存关系解析不特殊解析并列关系,识别并列关系较为困难.为此提出一种日语并列关系解析方法:基于支持向量机(SVMs)识别并列关键字,如果存在并列关键字,基于SVMs解析句中的并列关系,并将并列关系解析结果用于随后的依存关系解析.用京都大学语料库进行测试,结果表明并列关系解析与依存关系解析均取得较高的正确率.

Japanese parallel relation analysis based on support vector machines

Japanese parallel relation is one kind of Japanese dependency relation.It is difficult to recognize the parallel relation since conventional Japanese dependency parsers do not parse parallel relation in particular.A method of Japanese parallel relation analysis is proposed.The proposed model identifies parallel key based on support vector machines(SVMs).If there is a parallel key in a sentence,the parallel relation is parsed based on SVMs.The parallel relation is also used in the dependency analysis later.Experiments using the Kyoto University Corpus show that the system can obtain higher precision in Japanese parallel analysis and dependency analysis.