An analysis of web services attributes for discovery support

Abstract

The number of Web Services, available over the Internet, has increased due the Web Services properties that allow them to be reused and to develop loosely coupled systems. Nevertheless, the current number of Services complicates the discovery, that is why it is necessary to improve the current mechanisms, matching and similarity, that support locating them. This approach focuses in analysing the Web Services attributes, which are contained in their descriptions. Our aim is to con tribute to identify the role of attributes at the measurement of structural similarity function between web services and, through clustering, to develop resources for classifying and improving discovery. This proposal was tested through experiments over a collection of Service descriptions in WSDL which can be taken as an standard referent for such experiments. We show that our proposal outperforms the baseline taken as the best F−measure when the collection is represented by any attribute.