Using Recommender Systems to Assist in Intelligent Service Composition

Template-based composition of web services allows users to define templates with specified activities and to construct the control flow among the activities. And appropriate web services can be discovered and selected for each activity. However, choosing compatible services requires precise formal specifications of activities and templates because supporting systems lack "memory" of previous cases of using certain services together. In this paper, we propose to alleviate this problem by combining a template-based service composition framework with a recommender system and intelligent services. The combination not only supports users in defining the templates but also resolves issues of compatibility with limited computational power and less formal specification of templates using users\´ feedback from previous usage and services\´ feedback from intelligent services\´ communication ability.