Research On Semantic Retrieval Model Based On Domain Ontology

The traditional information retrieval is conducted only from the perspective of keywords matching,lacking of the semantic description of knowledge.With such retrieval method,the results returned often fail to satisfy the users.The traditional information retrieval can not be a good solution for the "polysemy" and "synonymy".In order to overcome the shortcomings of traditional information retrieval,semantic retrieval has become a new research direction in the field of information retrieval.In this thesis,we focus on the semantic retrieval model based on domain ontology,the main work is as follows.In this thesis,domain ontology is constructed.The principle and method of constructing domain ontology are analyzed and summarized,and the method of constructing domain ontology is summarized like in the process of software life cycle.Taking traffic domain for an example,the domain ontology is constructed through five stages: requirements analysis,design and construction,language description,effectiveness evaluation,model maintenance.Finally,the formal method is used to describe the classes and properties of traffic domain ontology.In this thesis,an improved algorithm of concept semantic similarity computation based on domain ontology is proposed.In this thesis,that the computing result of node density may be greater than 1 is improved,and the weight of each grandfather node is added on the basis of semantic coincidence degree.Considering the semantic distance,node depth,node density and semantic coincidence degree,an improved computational algorithm for conceptual semantic similarity based on domain ontology is proposed,and it shows that the improved algorithm is more accurate than the original algorithm by comparing and analyzing the results of the experiment.Finally,the concept semantic similarity and word concept relevance are combined to complete the mapping of words to concepts.In this thesis,the semantic retrieval model of domain ontology is constructed.Semantic feature vector is used to express the semantic of words,and semantic similarity is also calculated.It also analyzes the role and relationship of each module in retrieval model of domain ontology and constructs semantic retrieval model based on domain ontology.In this thesis,the domain ontology semantic retrieval prototype system is realized.In order to verify the validity of the semantic retrieval model based on domain ontology,the domain ontology semantic retrieval prototype system is realized by analyzing the design idea and workflow of the domain ontology semantic retrieval prototype system.According to the comparison results of the system operation,it shows that the semantic retrieval model based on domain ontology has some advantages.