Compared with normal web services,information-providing services have unique features that have seldom been considered in existing research on the automatic composition of web services.In this paper,we present a simple,yet well-formed,semantic-based capability model for information-providing web services,which can express such features as not modifying the world state and not requiring all input/output parameters to be supplemented with class information as semantics.We then present a corresponding automatic composition method derived from a query rewriting algorithm,MiniCon,used in the data integration field.This method adequately utilizes previous features,enables primitive semantic reasoning,and can generate executable BPEL scripts automatically.Performance of the method is complementary to traditional search-based ones.Experiments on a typical type of problem show that the method is usable in practice. Compared with normal web services, information-providing services have unique features that have seldom been considered in existing research on the automatic composition of web services. This paper, we present a simple, yet well-formed, semantic-based capability model for information -providing web services, which can express such features as not modifying the world state and not requiring all input / output parameters to be supplemented with class information as semantics. Then then a corresponding automatic composition method derived from a query rewriting algorithm, MiniCon, used in the data integration field. This method adequately accounts for previous features, enabling primitive semantic reasoning, and can generate executable BPEL scripts automatically. Performance of the method is complementary to traditional search-based ones. Experiments on a typical type of problem show that the method is usable in practice.