基于语义关联的实例相似度计算方法及应用研究

相似度计算方法的优劣直接影响到信息检索与推荐的效果。本文根据本体图模型中属性序列的特点,综合考虑层次关系和属性关系,在分析路径关联相似度、层次相交关联相似度及属性相交关联相似度的影响因素的基础上给出了实例之间综合语义相似度的计算方法。文章最后讨论了该相似度计算方法在解决基于内容的推荐中的过于专门化问题、协同过滤推荐中的稀疏性问题以及检索中查全率和查准率问题中的应用。

Research on Computation Method of Instance Similarity based on Semantic Association and its Application

The effect of information retrieval and recommendation is relative to the similarity computation method.Taking hierarchy relationships and property relationships into account, this paper analyzed the influencing factors of path association similarity, hierarchy join association similarity and property join association similarity according to the characteristics of property sequence in ontology diagram model, and then proposed a similarity computation method between instances.Finally, this paper discussed the applications of this similarity computation method in solving overspecialization problem of content-based recommendation, sparsity problem of collaborative filtering recommendation and recall ratio and precision ratio problems of information retrieval.