Research On Multi-Strategy Ontology Mapping Approache In Information Integration System

Information integration can shield the characteristics of data such as semi-structured, heterogeneity and distribution, and consequently provides a unified pattern for users to exchange information for heterogeneous data repository and to obtain valuable information from heterogeneous data repository.In information integration of heterogeneous sources it is a problem to solve semantic heterogeneity. Ontology is a clear and formal specification of sharing conceptual model, and it can effectively express the universal Knowledge in specific areas and can be used as the common semantic model of information integration system. Therefore, ontology-based data integration might be used to solve the problem of semantic heterogeneity.In ontology-based information integration, semantic conflicts could appear in much ontology, and it is the problem for ontology mapping to solve. At present, it is studied that ontology mapping could be discovered by calculating the semantic similarity of different concepts. The similarity calculating affects the precision of mapping. So calculation of semantic similarity is an important issue. Meanwhile, in existing approaches, similarity often derives from the instance, definition and structure of concept and etc. After research we find that the above three referenced objects reflect the relations between concepts in some degree. Therefore, this paper uses a multi-strategy ontology mapping system, considering the various characteristics of ontology, including the instances, definition and structure of concept and so on. This system can improve the accuracy of ontology mapping, and complete mapping mission better.Most of the current mapping methods have not made the most of the mapping history. This paper proposes a mapping reuse method of taking full advantage of mapping experience to discover the potential mapping, and improves the mapping efficiency. However, experiments show that the multi-strategy does not always outperform single-strategy. We propose a detecting approach of multiple strategies. So far the results obtained show that multi-strategy detection improves precision and recall significantly. In brief, we make the following contributions in this paper:We propose multi-strategy ontology mapping, introduce the composition of the mapping system briefly. This mapping system utilizes of various ontology information ,so it effectively improve the accuracy of the mapping; We propose the model of the ontology- based information integration as a test platform for mapping system; We describe the mapping reuse algorithm in detail and talk about how to storage the mapping history. Mapping reuse improves the efficiency of ontology mapping; In order to detect the optimal set of strategies we have improved a detection algorithm, and it improves the efficiency of the detection. We introduce how to calculate the semantic similarity of the ontology using multi-strategy ontology mapping system.