基于语料库的关联词识别方法

The Method of Identifying Discourse Maker Based on Corpus

关联词语的出现在汉语文章中占有很大的比重,它们可以在复句分析中起到非常重要的作用。该文主要讨论如何在语料库中提取反映关联词语用法的各种特征信息,再通过数据挖掘算法将这些信息整理为关联词语的知识库,作为关联词识别的依据,并对有歧义的关联词语提出了基于决策树的消歧方法。最后,通过开放测试证明了该方法的有效性。

With the high occurrence rate in Chinese articles,discourse maker plays a significant role in automatic processing of sentences.This article elaborates on how to obtain characteristic information for discourse maker from corpus,and builds knowledge base with data mining algorithm as the basis for discourse maker identification.Also it proposes the method of identifying ambiguous discourse maker with decision-tree.At last,the validity of the method is proved by test.