Genetic Algorithm Based WEB Service Selection With Multiple Constraints And Preferences

With the development of Web services, the number of Web services with same function while different QoS attributes is growing. Facing with lots of combination schemes, Web services selection based on overall QoS restrictions, plays a very important role in Web service combination. Users demand much more about the QoS of Web services Combination, and sometimes users need to generate self-satisfactory Web services according to user preference area. So it becomes an important research subject that how to provide solutions to meet overall QoS constraint and user preference. The specific content of this paper is as follows.Firstly, this paper studies Web service selection based on QoS constraint. This paper proposes a combination service QoS calculating method to support services associated constraints. We also analyse the constraint boundary and the feasibility measures of individual, propose a double archive model with the feasible solutions and infeasible solutions in the evolution. According to the model, we propose a double archive algorithm, the feasible solutions and infeasible solutions use different selection methods. The feasible solutions use fast non-dominated sort algorithm and the crowding distance. The infeasible solutions select the best infeasible solution according to the fitness, and then find the real Pareto optimal solution according the simulated binary crossover operation and mutation operation.Secondly, the Web service selection based on preference is studied. We propose the preference of the QoS weight and user preference model of Web services. According to the model, we propose the level of preferred method based on preference information and the positive and negative reference point preference algorithm. We sort the level of the schemes according to the fitness value, algorithm uses binary tournament method, if they have the same level, we use the positive and negative reference point operator to select close to the positive reference point, keep away from the negative reference point, making the algorithm produce more feasible solutions in the interest area of users.Finally, we verify the effectiveness of the algorithm through the experimental results. We compare the algorithms and apply the algorithm to the Web service optimum selection to get the ideal Pareto optimal solutions.