基于语义相似度的个性化信息检索方法

SEMANTIC SIMILARITY-BASED PERSONALISED INFORMATION RETRIEVAL METHOD

随着网络资源的爆炸式增长以及语义网的出现,个性化检索成为当前信息检索研究的热点。提出一种基于语义相似度的个性化信息检索方法。通过分析三种传统的语义相似度计算方法,针对三种方法的优缺点,提出一种改进的基于领域本体的语义相似度计算方法,该方法将有向边权重、概念层次差以及概念属性,引用到语义相似度的计算中。并在已有的领域本体的基础上构建用户个性化模型,使用本体来存储用户兴趣,最终应用到个性化检索中。实验结果表明,该方法有效地对用户的查询请求进行概念扩充,提高了搜索的查全率与查准率。

With the explosive growth of web resource and the emergence of semantic web,personalised retrieval becomes the focus of current information retrieval research.In this paper,a semantic similarity-based personalised information retrieval method is proposed.According to advantages and disadvantages of three traditional semantic similarity calculation algorithms analysed,an improved semantic similarity calculation algorithm method based on domain ontology is put forward,it introduces the directed edge weight,concept hierarchy difference and concept property into the computation of semantic similarity.Moreover,based on existing domain ontology,the personalised user＇s model is constructed,in which the user＇s interest is stored in ontology,and at last this is applied to personalised information retrieval.Experiment results show that the user＇s inquiry request has been expanded its concept effectively,and the recall and accuracy of retrieval have been improved obviously.