In the recent years, Web-service has greatly changed the way people use the Internet.How to take advantage of the web services effectively and efficiently is one of the mostimportant issue. Most of the studies before are based on very simple or artificial test sets,which is not so persuasive on both practicality and social relations.In order to give the paper a strong experimental support, a corpus was gained fromthe Internet, including5537WSDL files. After parsing WSDL files, informationpreprocessing, natural language processing, ontology matching and automatic semanticannotation,4web-service networks were built with different granularity levels. Andbased on the web-service networks,3important web-service social relations: Substitution,Competiton and Collaboration were proposed. In the expertiment, the proposed methodreached a single operation recommendating accuracy of average higher than90%; and forthe natural language fitting recommendating, the accuracy is88.2%and more than70%recommended web service operations are obviously useful. |