QoS-driven selection of web service considering group preference

The aim of this paper is to present an approach for selecting Web services during the process of building an orchestration/composition, on the basis of some Quality of service (QoS) features. The specific proposal is to consider decision maker's preferences, and aims at guaranteeing consistency of the selection considering such preferences. The composite service has multiple tasks and each task can be fulfilled by a group of services. Then the optimization is to select a service for each task so that the overall evaluation function can be optimized. With the traditional group decision method, the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) process, the decision makers must construct the decision matrix. Once web service candidates become unavailable or new web services emerge, the alternatives change, which means that the decision makers must reset the large number of evaluated values. In this paper, a group selection process named Group Decision Web Service Selection, GDWSS, was proposed. First, weights for each task are collected from the different decision makers and global optimization parameters are generated. Next, an easy preference setting named WSS-TOPSIS, adapted from the TOPSIS method, is proposed to reduce the amount of preference data input. Furthermore, GDWSS allows group decision makers to set preferences with a lighter workload and to consider group preferences for the system to reduce the exception difference existing among decision makers. Finally, the scenarios are replicated to illustrate the effectiveness of this approach.