A QoS-aware Service Composition Approach based on Semantic Annotations and Integer Programming

Purpose – The purpose of this paper is to propose a novel quality of service (QoS)-aware service composition approach, called SEQOIA, capable of defining at run-time a service composition plan meeting both functional and non-functional constraints and optimizing the overall quality of service. Design/methodology/approach – SEQOIA is a semantic-driven QoS-aware dynamic composition approach leveraging on an integer linear programming technique (ILP). It exploits the expressiveness of an ontology-based service profile model handling structural and semantic properties of service descriptions. It represents the service composition problem as a set of functional and non-functional constraints and an objective function. Findings – The authors developed a proof of concept implementing SEQOIA, as well as an alternative composition solution based on state-of-the-art AI planning and ILP techniques. Results of testing activities show that SEQOIA performs better than the alternative solution over a limited set of candidate services. This behaviour was expected, as SEQOIA guarantees to find the service composition providing the optimal QoS value, while the alternative approach does not provide this guarantee, as it handles separately the specification of the functional service composition flow and the QoS-based service selection step. Originality/value – SEQOIA leverages on semantic annotations in order to make service composition feasible by coping with syntactic and structural differences typically existing across different, even similar, service implementations. To ease the adoption of SEQOIA in real enterprise scenarios, the authors chose to leverage on an XML-based message model of services interfaces (including but not strictly requiring the use of WSDL).