一种基于同义词消除的文本过滤方法

传统的向量空间过滤模型通常是提取字、词、短语等作为特征项,这样做的缺点是没有考虑文本的语义信息。文章提出了利用知网对向量空间模型的文本向量进行同义词消除的过滤方法。该方法比传统的单纯基于关键词匹配的方法更精确地体现了文本之间的相似度,提高了过滤性能,同时也降低了向量空间的维数,减少了计算量,提高了过滤的效率,实验结果表明基于该文的过滤方法确实提高了系统的性能。

A Text Filtering Method Based on Synonym Elimination

The traditional VSM(Vector Space Model) system often selects words as the feature,and one of its shortness is the lack of semantic information in document representation.In this paper,we present a kind of filtering method which selects all of the synonyms of each feature in the document vector and replace them as one of them based on HowNet.This method approves the system's efficiency and precision,and the result of the experiment also proves its availability.