The Reasch Of Ontology Concept Mapping Based On Machine Learning

Ontology is considered as an important theoretical foundation in the fields of information integration, Semantic Web and knowledge management. However, in many cases, the same concept may be interpreted into diverse ontologies in different fields, and sometimes even different institutions in the same domain may define the same concept with different ontologies. Thus, problems concerning heterogeneous ontologies naturally arise. To solve the problem and achieve interoperability among different ontologies, ontology mapping is the best way out. That is to build a mapping bridge among ontologies so as to realize knowledge sharing and integration of ontologies.Based on the existing methods on ontology mapping, this paper intends to analyze their characteristics, find out their problems and improve their methods of similarity algorithm on ontology mapping so as to enhance interoperability and adaptability of ontology mapping as well as the comprehensiveness of similarity algorithm. And finally a framework of ontology mapping is designed on the basis of this algorithm.Giving full consideration to the factors that influence similarity algorithm as instances of concepts, hierarchical relationship and their attributes, the author designs a computational model based on the integrated semantic similarity. The similarity of ontology concepts is calculated from the perspective of the name, attribute and instance of the concept, and then integrates these three similarity conponets and weights to obtain the integrated semantic similarity of concept. The originality of this paper lies in its introduction of machine learning techniques to semantic similarity algorithm. The ontology concept mapping based on machine learning has a better semantic match, which has been applied in the fruit field and achieves an objective effect. It also offers a solution to achieve semantic interoperability of the information in the environment of heterogeneous ontologies.