Efficient Optimized Composition of Semantic Web Services Rattikorn Hewett, Bach Nguyen and Phongphun Kijsanayothin Department of Computer Science, Texas Tech University Rattikorn.Hewett@ttu.edu, Bach.Nguyen@ttu.edu, kphongph@gmail.com Abstract— Advances in Internet and software technology play increasing roles in creating work environments and online services that impact our everyday living. Web service computing paradigm has revolutionized how these software applications can be developed rapidly and reliably by employing available serv- ices available in standard forms across the web. Automatic com- position of web services is necessary and challenging when there are a huge growing number of available web services but no single service offers the desired function. This paper presents an approach to automatically combining published web services to meet the application requirements using as few services as possi- ble. Our approach is based on heuristic algorithms that incorpo- rate semantics of services from their ontology into identifying valid executable services in the composition structure. The paper describes the approach and evaluates its performance on public service repositories. Our experimental results show correct op- timized solutions 100% of the time, with a reduction in the aver- age running time, compared to champion of Web service chal- lenge’06 competition, of 60.2% over 31 composition problems. Keywords: semantic web services, automatic service compo- sition, service-oriented computing, ontology, heuristic search. I. INTRODUCTION Advances in Internet and software technology play in- creasing roles in creating work environments and online serv- ices that impact our everyday living. It is no longer necessary to go to bookstores to buy books or to visit banks to complete certain transactions. Many types of services can be obtained any time and anywhere that are networked accessible across the Internet. Modern enterprises rely on software to conduct business or perform day-to-day operations. Web service computing paradigm has revolutionized how developers can rapidly and reliably create these service-oriented software applications by employing available web services. The term web services (or services) refer to self-contained platform independent software units prescribed by standard machine- readable specifications and protocols to support interoperable services for distributed applications across the Web [4, 6]. When there is a huge growing number of available web services that no single service can offer the desired function, automatic composition of web services becomes necessary and challenging. Current approaches to automatic web serv- ice composition in the literature (see [1, 4, 6, 7] for surveys) often include service discovery techniques for selecting ap- propriate services and integrating them into a composite serv- ice structure that has been pre-specified [7]. In syntactic composition, syntactical languages such as BPEL and WS- CDL are used to specify interfaces and process flows of com- bined services, while semantic composition relies on DAMLS (ServiceModel), for specifying semantics of service processes along with service ontology languages (e.g., OWL-S and WSMO) for guiding inferences in planning systems (e.g., Golog, HTN) [3, 6]. However, most of these approaches re- quire manually written templates of composite structures. No assembling of complex flows from atomic service message exchanges is automatically derived from a search process [7]. Other automated approaches to web service composition are based on formal methods including logical inductions and model checking for deriving composition paths [4]. However, they tend to be very complex and computationally expensive. Furthermore, service compositions based on rules, or logical planners [3, 4, 7, 8] often rely on representational models that may be suitable for planning problems but unnecessarily complex for service composition problems. This paper addresses the issue of semantic web service composition, specifically how to efficiently and automatically construct a valid executable composition (sequence) of pub- lished services to satisfy the application requirements with minimal number of services. Our approach is based on an effective modeling approach and two heuristic algorithms that incorporate semantics of services from their ontology into identifying valid executable services in the composition struc- ture. The rest of the paper is organized as follows. Section II describes related work including champion of Web service challenge’06 competition [10] system by Weise et al. [9], which we use in our comparison study. Section III defines the problem and introduces our modeling approach. Section IV presents our composition approach along with its complexity analysis and Section V illustrates its use for solving service composition problems. Section VI empirically evaluates the approach by comparing our results with those obtained by Weise et al.’s system. The paper concludes in Section VII. II. RELATED WORK Several automated approaches to semantic service compo- sition have been proposed [3, 5-9, 11, 12]. Most use back- ward search control, like in planning systems, and indexing mechanisms to improve efficiency. The optimization of the number of services deployed in a composition has been stud- ied using graph-based [9] and planning-based [5] approaches. Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics San Antonio, TX, USA - October 2009 978-1-4244-2794-9/09/$25.00 ©2009 IEEE 4165