ABSIM: AN AUTOMATED BUSINESS SERVICE IDENTIFICATION METHOD

Abstract

One of the key activities in the service-oriented solution is the identification of services according to a set of predefined design principles. Existing service identification approaches are often prescriptive and are based on the architect’s experience, and therefore might lead to non-optimal designs which result in lower performance, reduced scalability, and complicated dependencies between services. In this paper, an automated method, called ABSIM, is proposed to identify business services by adopting design metrics based on top-down decomposition of business processes. ABSIM first gets a set of business goals along with business processes as its input and then produces a set of non-dominant solutions using a multi-objective genetic algorithm. Since each produced non-dominant solution corresponds to a set of services, in the next step of the method, fuzzy logic is used to rank the obtained solutions according to the business goals specified at the first stage. In this way, using ABSIM, a set of services which are based on the enterprise business processes and goals and are appropriate in terms of qualitative attributes is obtained. We use two case studies to show the applicability of ABSIM. Also, we demonstrate the appropriateness of ABSIM results through comparing them with outputs of expert’s experiences based on a proposed formal approach.