基于语料库的语义解释空间构建及其应用

Construction and application of the lexis semantic interpretation space based on corpus

在汉语中,词汇是研究语义的基本单位.词汇不是孤立存在的,它们之间存在着各种各样的联系,其中语义联系是一种重要的联系,而语义距离则是语义联系的定量表示.为了得到词汇之间的语义距离,文中提出基于语料库建立语义解释空间的方法,该方法把词汇映射为空间中的点,并给出其坐标,求出关联词汇的几何距离.语料库中的循环解释利用拓扑排序方法检测并进行处理.在此基础上,构建语义解释空间原型系统,并应用于语义关联词汇检索和试卷中主观题的评分辅助矫正.该空间的构建,为进一步研究词汇之间的跳跃转换等提供了研究基础,也使得把句子映射为空间中的曲线,篇章映射为空间中的曲面成为可能.

Word is the semantic unit of Chinese language. It is sure that words do not exist in isolation and there are many relationship among them and one of the important relationship is semantic connection,for which a quantitive index is the semantic distance. In order to get the semantic distance among words,the method to construct the lexis semantic interpretation space（LSIS） based on corpus is studied,by which,the words are mapped to the points in the space,and all the points have their own coordinates. Then based on the coordinates,the distance among words can be calculated. The circular explanation in corpus is dealt with topological sorting.A LSIS prototype system is given and two application are introduced： semantic relative-words retrieving and LSIS aided correction of scores of subjective questions. The construction of LSIS supports the basis to shift from words to words and give a method to map sentences to curve and documents to surface.