The Design And Implementation Of An Ontology Matching System Based On Multi-strategy

Semantic Web gives definite semantic meaning to the information and it facilitate the collaborative work between human and computers. Ontology is the core of Semantic Web and it is a formal explicit specification of a shared conceptualization including a wealth of natural language descriptions. With the development of semantic Web, more and more systems begin to use ontologies to describe Web information. However, the heterogeneous ontologies are the obstacles for semantic interoperation between different systems. In order to make more effective use of knowledge in the Semantic Web and achieve semantic interoperability between different but related ontologies, we need ontology matching to build the interaction rules between heterogeneous ontologies. Although many researchers have proposed novel approaches to match ontologies, the results of the existing ontology matching system remains to be further improved.Based on those facts, this thesis focuses on the research of ontology matching algorithms and we put forward an ontology matching algorithm based on multi-strategy which means matching two ontologies using many matching algorithms and then implement the algorithm as a system. Specifically, the main contributions of this thesis are listed as follows:(1) Making a research on different implementations of multi-strategy matching. The multi-strategy matching could be implemented by using similarity integration or classify algorithms based on machine learning. The research focuses on the pros and cons of different implementations of multi-strategy matching.(2) Putting forward an ontology matching algorithm based on multi-strategy. The linguistic matching stage of the algorithm use the multi-strategy method which is implemented by KNN classification algorithm. At structural matching stage we propose an algorithm based on GEL graph. Cadidate matching set is generated by adjacent nodes of the matches that have been found and then we get the final match set using multi-strategy matching based on KNN. We also put forward the corresponding methods to detect and repair the inconsistent mappings.(3) Implementing our algorithms as an ontology matching system and comparing our system with several ontology matching systems. Our system obtained better precision and F1 value.