摘要
兼语句是文本知识中一种较为常见又比较特殊的句式,对兼语句进行知识获取方面的研究是文本知识获取的一个重要研究方向。为了构建一种新的兼语分类体系,从句中第一个谓词的角度出发将兼语句式分为八个大类,并在语义分类和描述框架的基础上对这八个大类进行进一步细分;然后从兼语中第二个谓词发生的时序角度出发进行归纳分类;最后,对于不能充当兼语句式中第一个谓词成分的语义类,从语义类的层级上分析总结了其原因和规律。该分类体系比已有的分类体系更全面更细致,它几乎涵盖了文本知识中所有的兼语句。实验结果表明,该分类体系在语料扩充上正确率达到97.78%,是有效可行的。
Subjective-object structure which is also called the subjective-object sentence is a special and common kind of sentence structure in the text knowledge. The research of knowledge acquisition in subjective-object structure is an important research direction of text knowledge acquisition. In order to build a new classification system of subjective-object structure, firstly, this paper classified the first predicate in the subjective-object sentences into eight categories on the basis of the framework of semantic taxonomy and description（FSTD）. Then it made an induction and classification of the second predicates from the perspective of time sequence. Finally,this paper analyzed the reason why some of the semantic classes couldn＇ t act as the first predicate in the subjective-object sentences. The new classification system is more comprehensive and detailed than the existing researches. It covers almost all the subjective-object structure in text knowledge. Experiment results show that the classification system is effective and feasible which has the accuracy of 97.78% in the corpus expansion.
出处
《计算机应用研究》
CSCD
北大核心
2017年第1期15-20,共6页
Application Research of Computers
基金
国家自然科学基金资助项目(91224006
61035004
61173063
61203284)
国家科技部资助项目(201303107)
关键词
兼语句式
语义分类
语义组合
subjective-object structure
semantic classification
semantic combination
