Collecting, annotating, and classifying public web services

Abstract

The limitations of the traditional SOA operational model, such as the lack of rich service descriptions, weaken the role of service registries. Their removal from the model violates the basic principles of SOA, namely, dynamic binding and loose coupling. Currently, most service providers publish their Web Services on their websites instead of publishing them in service registries. This results in poor usability of these Web Services especially wrt. service discovery and service composition. To handle this problem, we propose to increase the usability of public Web Services by collecting them automatically from the websites of their providers with the help of web crawling techniques. Additionally, the collected services are annotated with descriptions that are extracted from the crawled web pages and tags that are generated from the same web pages. These annotations are then used to derive a classification for each Web Service into different application domains. In this paper, we introduce the details of our approach and show its practical feasibility through several evaluation experiments. © Springer-Verlag 2010.