Correlation-aware constrained many-objective service composition in crowdsourcing design

Abstract

With the development of social productivity, crowdsourcing design, a new service-oriented innovative design model emerged. The optimal selection of service resources in crowdsourcing design has a great impact on the quality of task implementation and consumer satisfaction. Unlike the previous research, this paper models and solves the problem from the perspective of constrained many-objective optimization. Firstly, a constrained 6-objective combination optimization model is constructed according to the characteristics of service resources in crowdsourcing design. Secondly, a variety of correlation categories and correlation forms of services are considered. It is not limited to the correlation between two services but rather correlations between any number of services. In order to improve search efficiency, a multi-objective service pruning (MOSP) method based on efficient non-dominated sorting (ENS) is used to reduce the decision space. The new constraint strategy handles complex constraints, and a correlation-aware local search strategy is proposed to deal with correlations between services. The results of comparison experiments prove the superiority of the proposed method in solving the constrained correlation-aware many-objective service composition problem.