一种基于本体语义的信息检索模型

对基于向量空间模型的检索方法进行改进,提出基于本体语义的信息检索模型。将WordNet词典作为参照本体来计算概念之间的语义相似度,依据查询中标引项之间的相似度,对查询向量中的标引项进行权值调整,并参照Word-Net本体对标引项进行同义和上下位扩展,在此基础上定义查询与文档间的相似度。与传统的基于词形的信息检索方法相比,该方法可以提高语义层面上的检索精度。

A Information Retrieval Model Based on Semantic of Ontology

The information retrieval model based on semantic of ontology was proposed by making improvements on traditional vector space model.Firstly,similarity between concepts was computed by taking WordNet lexicon as referencing ontology.Secondly,the weights of terms in query vectors were adjusted according to the similarity among them.Meanwhile,the terms in query vector were expanded by using synonymy and hypernymy/hyponymy associations occurred in WordNet ontology.Finally,the similarity between query and document was computed.Compared with traditional information retrieval method based on word forms,our method can improve the retrieval precision on semantic level.