Abstract: We propose an ontology-driven discovering model of geographical information services to improve the recall and the precision of geographical information services,study the multilevel matching arithmetic of geographical information services.This arithmetic filters and matches the services in service register center according as the similarity between services selected with services requested through defining the function similarity and credit standing similarity.The matching arithmetic,geographical information services ontology,and semantic describing form the discovering model.The results show that the ontology-driven discovering model is excellent in recall and precision,and can maintain dynamic loading balance of service copy.