Design And Implementation Of Ontology Matching System Based On Domain Background Knowledge Base

Ontology matching is one of important techniques as well as methods for resolving semantic heterogeneity,and meanwhile it is also one of research focuses for semantic integration.With the development of Linked Open Data,some huge open datasets provide a lot of concepts and knowledge structure in various domains.If domain-specific knowledge existing in these huge datasets can be used in ontology matching,it is helpful to improve the effects of ontology matching of specific domain.Hence,in order to realize the ontology matching in metallic materials domain,in this paper,we propose a matching strategy based on background knowledge base.In addition,we also design an approach to extract metallic materials knowledge from Yago so as to construct a background knowledge base.The research work is as follow.1)For the existing open datasets(i.e.,Yago),an approach is designed to extract domain-specific knowledge based on the structure of Yago and string matching algorithm.First,we define keywords to initially obtain some domain-specific concepts from Yago by string matching algorithm.Then,based on these obtained concepts,the domain-specific knowledge and related knowledge is extracted according to the structural features of Yago,so as to acquire a domain-specific ontology.Finally,a strategy is designed to construct a background knowledge base according to the knowledge extracted from Yago.2)The matching strategy and the matching methods are designed according to the features of the background knowledge base.First,the concepts(i.e.,candidate concepts)from input ontology should be found the corresponding concepts in domain background knowledge base.In other words,a strategy is designed to find the target concepts in domain background knowledge base.Then,based on the candidate concepts and the structure of domain background knowledge base,we build the sense group of candidate concept,and meanwhile design the strategy and method to compare the sense groups so as to get the matching scores among candidate concepts.Finally,the relation between two candidate concepts could be defined according to the matching score.3)With the background of metallic materials domain,we evaluate and analyze the validity of the matching method by experiments.In addition,extraction approach of domain-specific knowledge,construction strategy of knowledge base and matching methods for metallic materials domain are evaluated inlight of F1-measure and time performance,and the experimental results demonstrate that the proposed approaches are feasible and the time performance is acceptable.Meanwhile,a prototype system is developed to visually demonstrate the knowledge structure of MMOY and background knowledge base.A ontology matching system is also designed to show the whole matching process.