基于概念关联度的智能检索研究

Research of intelligent retrieval based on concept correlation degrees

为提高智能检索系统的查准率和查全率,分析了目前查询扩展方法存在的不足,考虑计算有向边权重的因子对语义距离的影响,对基于语义相似度的计算方法进行优化,提出了一个计算用户查询与文档相关性之间关联度的表达式。利用构建的领域本体量化概念间的关联程度,通过计算合理的相似度与相关度的权重来进行查询扩展,并设计了一个基于概念关联度的语义检索模型,将检索结果按关联度排序显示。实验表明,该方法在确保查准率的前提下能有效提高查全率。

To improve the precision ratio and recall ratio of the intelligent retrieval system, the present problems of query expansion method are analyzed. Thinking the influence of directed edge weight factor to semantic distance, optimizing the semantic similarity calculation method, a correlation degree expression between user query and document relevance is proposed. This method quantifiy correlation degree between concepts on the conducted ontology. User query is to expand through the calculation of reasonable similarity and relevance weights. A semantic retrieval model based on concepts correlation degree is designed. And the result of every search is displayed in order by correlation degrees. Finally, the experimental results show that the method can raise the recall rate effectively under ensuring the precision rate.