一种基于本体论和潜在语义索引的文本语义处理方法*

提出一种基于潜在语义索引和本体论的文本语义处理方法。首先构建一个基于本体论的虚拟标准文本特征向量，然后采用潜在语义索引方法以虚拟标准文本特征向量为参照对文本集进行语义聚类，最后在虚拟标准文本特征向量的导引下利用本体库中的知识对聚类获得的文本集合的类别和语义进行显性标注。实验表明，该方法能较好地在语义层面对文本进行有效的聚类，而且聚类结果能显性地显示类聚所属的类别。

A Text Semantic Information Processing Method Based on Ontology and Latent Semantic Indexing

This paper presents a method for text semantic information processing based on Ontology and latent semantic indexing. Firstly, virtual standard text characteristic vectors are constructed; then, the texts are semantically classified into document sets according to virtual standard text characteristic vectors by using latent semantic indexing method ; finally, semantically explicit annotations to the document sets are abtained from Ontology - base by guidance of virtual standard text characteristic vectors. Experiments show that method can achieve good text clustering of semantic level, and the clustering can explicitly indicate categories of the clustered documents.