Abstract: Over the past few years, semantic Web and grid computing have developed separately in distinct communities.The semantic grid, which is the cross of two fields, is the emerging research focus recently.Ontology can enhance the self-organization of grid by applying the semantic layer on the grid.Ontology is a good solution for interoperation inside a grid community.However, current grid community is using a centralized, persistent, scalable ontology repository.Ontology heterogeneity among grid communities is becoming an ever more important issue.In this paper, a multi-strategy learning approach is proposed to resolve the problem.An ontology mapping system is described, which applies multiple classification methods to learn the matching between ontologies.It uses the general statistic classification method to discover category features in data instances and use the first-order learning algorithm FOIL to exploit the semantic relations among data instances.On the basis of the prediction results of individual methods, the system combines their outcomes using the matching committee rule called the Best Outstanding Champion.The experiments show that the system achieves high accuracy in the real-world domain.