A User-Preference Driven Lexicographic Approach for Multi-Objective Distributed Web Service Composition

In distributed Data-intensive Web Service Composition (DWSC), multiple individual services from different locations are composed to accomplish a more complex service. Automatic DWSC must fulfil functional requirements and optimise multiple Quality of Service (QoS) attributes simultaneously. Users often have QoS preferences, e.g. one QoS attribute is deemed more important than the others. Users' QoS preferences are crucial parts of a service composition request. However, user preferences have not been considered in the current composition research. This is mainly due to the difficulties of accurately representing user preferences during the optimisation process. To model users preferences easily, we present them through the lexicographic ordering. We then develop an effective multi-objective evolutionary algorithm, i.e. Lex-NSGA-II, which uses a newly designed clustering technique to produce high-quality solutions that satisfy users' QoS preferences. Our experimental evaluations using benchmark datasets demonstrate the effectiveness of our proposed method.