基于领域本体模型的概念语义相似度计算研究

Research on Concept Semantic Similarity Computation Based on Domain Ontology Model

随着本体在信息检索、人工智能等领域的广泛应用,面向本体的概念相似度计算成为本体研究的一大热点。目前领域本体中概念相似度的研究主要是利用概念的上下位关系进行计算,但这并没有完整反映出概念的语义信息。本文首先在本体模型的基础上提出领域本体模型的八元组表示方法和领域本体概念的九元组表示方法,给出领域本体模型的DCG图。然后以提出领域本体模型为基础构建概念语义相似度计算的MD4模型,该模型全面考虑概念的属性、上下位语义结构关系、自定义语义关系和实例特征对相似度的影响,通过综合计算,得到领域本体中概念的实际相似度。本文最后以动车组专业本体作为实验对象,对MD4模型进行验证。实验结果表明,该计算模型充分利用概念的语义信息,得到的结果也比较合理。

Along with the widespread application of ontology in the fields of information retrieval and artificial intelligence etc,the concept similarity computation of domain ontology has become the focus research field.Currently,most research on concept similarity computation is based on ＂is a＂ relation between concepts,however,it does not utilize the concept semantic information completely.Firstly,on the basis of ontology model,this paper proposes an Eight-tuples model for domain ontology and a Nine-tuples model for its concept.We can use the Directed Cyclic Graph（DCG） to formalize the domain ontology.Then this paper builds the fourfold Matching-Distance Model（MD4） to compute concept semantic similarity based the on domain ontology model.This MD4 model has comprehensive consideration about the property,＂is-a＂ relation,＂user-defined＂ relation and instance of concept.And through integration computation,we can get the true concept similarity.Finally,this paper makes an experiment on the EMU ontology that was built in the high-speed railway domain.The experiment shows that the method utilizes the concept semantic information fully and the computation result is reasonable.