Ranking web service for high quality by applying improved Entropy-TOPSIS method

Abstract

With the increasing numbers of web services published on the Internet, there are many services existed that fulfill user's requirements equally well even after filtering by functional requirements, so how to rank the similar services based on QoS becomes an important issue. Due to the multi-dimensional attributes of QoS, it can be treated as a multiple criteria decision making (MCDM) problem. Since the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) is widely used in dealing with MCDM, we propose a novel ranking approach based on TOPSIS to solve the service ranking problem. Noticed that the weight of every dimension in alternatives will directly affect the ranking result in the TOPSIS method, determination of weight for each dimensional QoS attribute can be a key factor in service ranking. We use Shannon's Entropy theory to get the objective weight, which reveals the intrinsic impact from different QoS attributes to the overall service quality in actual. Then combine with subjective weight, to get the comprehensive weight, it can not only improve the objectivity of the weight-making process, but also better closer to the user's personalized needs. We have shown the applicability of the ranking approach by using a case study. It illustrates the Entropy-TOPSIS method for service ranking process, and the simulation results confirm the feasibility and validity of proposed method.