Improving web service discovery with usage data

Web service discovery is a difficult and challenging activity that makes the development of service-based applications a time-consuming and still not widely practiced process. Descriptions of publicly available services are scarce and their quality isn't guaranteed. This article presents a recommendation system to help developers of service-based applications discover and select appropriate services. Given a task description, the system recommends service operations according to the history of decisions previously made for similar objectives. The system is developed using IC-Service, a domain-independent recommendation Web service based on the implicit culture theory of service developers. IC-Service automatically collects information about service usage. Experimental results show that the system can learn from experience and achieve fair precision in its recommendations. This article is part of a special focus on service-centric software systems. © 2007 IEEE.

Service-oriented computing and Web services are becoming more popular, enabling organizations to use the Web as a market for selling their own services and consuming existing services from others. Nevertheless, the more services are available, the more difficult it becomes to find the most appropriate service for a specific application. Existing approaches to Web service discovery tend to address different information-processing styles. However, Web services have functional and nonfunctional characteristics that can be difficult to present and control. Service behavior and quality-of-service (QoS) parameters can vary over time, and new services can emerge in certain business areas.