The Research Of Multi-Strategy Based Ontology Mapping

In recent years, ontology as a domain knowledge conceptualization andmodeling approach has been widely recognized. However, the lack of a unifiedontology construction tools, making different organizations or individuals alwaysexploit ontology according to their own need. This has resulted in informationheterogeneity between the different ontologies. In order to better achieve theknowledge sharing and exchange, we need to establish semantic correspondencesbetween the different ontologies, and this is the work of ontology mapping.Ontology mapping based on multi-strategy uses a number of ways to achieve themapping between the ontologies. Each strategy analysis certain aspect of ontologyfeatures such as linguistics, structure. For this reason, in order to better achieve theontology mapping work, this paper starts with linguistic strategy and structuralstrategy, studying their shortcomings and improving. The main work of this paper isas follows:(1) Many methods are based on linguistic to achieve ontology mapping.One ofthem based semantic similarity calculation is widely accepted.This methodrepresented ontology concepts as feature vectors, so that calculation of the similaritybetween two concepts is translated into the calculation of similarity between twofeature vectors. Therefore, how to represent the feature vector of concept is a key insuch method. To address this issue, this paper proposes an ontology mapping methodbased on wikipedia(OMW). This method uses Wikipedia to semantic enrichment forevery concepts of ontologies. Thus the concepts are represented as feature vectors,Finally this method uses cosine measure to calculate similarity. This method is alsointroduced the idea of propagation, the feature vector of concept is modified to makeit a better representation of semantic information of concept. Experimental resultsshow that the proposed approach can achieve a better mapping task, whileovercoming the shortcoming of iMapper systems that the representation of featurevector over-relies on document collections.(2) The theoretical basis of ontology mapping based on structure to enhance isthat the concepts of ontologies are similar when their adjacent concepts are similar.Therefore, how to represent the similarity of their adjacent concepts is a key. Toaddress this issue, this paper proposes a method that uses the structure of the ontology graphs to enhance the accuracy of ontology mapping: Given a conceptual entity,computing the similarity of their ancestor, and computing the similarity of theirproperties based on the Domain relationship in OWL. Experimental results show thatthe proposed approach can extract new mappings, while superior to the traditionalmethods.