Abstract: The retrieval mismatch problem between the user queries with the knowledge of the document, put forward a body retrieval based on the similarity matrix matching algorithm. The method is an improvement on the existing search matching algorithm. The use of domain ontology to quickly determine the set of candidate matches; consolidated from three levels of similarity between elements; similarity matrix is the similarity between the two ontology. Experimental results show that it has a better performance in the recall, precision and response time.

Key words: ontology matching, knowledge retrieval, similar measure, classify

摘要： 针对用户查询与知识文档间检索失配问题，提出了一种基于相似度矩阵的本体检索匹配算法。该方法是对现有的检索匹配算法的一种改进。方法利用领域本体来快速确定候选匹配集；进而从三个层次来综合计算元素间的相似度；利用相似度矩阵得到两本体间的相似度。实验结果表明，在查全率、查准率和响应时间方面都有较好的表现。

关键词: 本体匹配, 知识检索, 相似度计算, 分类