Weighted fuzzy clustering for capability-driven service aggregation

Abstract

Workflow design, mashup configuration, and composite service formation are examples where the capabilities of multiple simple services combined achieve a complex functionality. In this paper we address the problem of limiting the number of required services that fulfill the required capabilities while exploiting the functional specialization of individual services. Our approach strikes a balance between finding one service that matches all required capabilities and having one service for each required capability. Specifically, we introduce a weighted fuzzy clustering algorithm that detects implicit service capability groups. The clustering algorithm considers capability importance and service fitness to support those capabilities. Evaluation based on a real world dataset successfully demonstrates the effectiveness and applicability for service aggregation.