Web Services Composition Optimization Research Based On Pareto Multi-objective Artificial Bee Colony Algorithm

Web Service is a new distributed computing model that can efficiently integrate information and data on the Internet. It is a new direction of integration technology. Due to more and more services with similar functions but different QoS, the search space of services increases largely, which makes services composition problem more complex.Based on this, the dissertation designs a Pareto multi-objective artificial bee colony algorithm to solve Web services composition optimization problem. The main work can be described as follows:Firstly, the dissertation proposes a construction method of Pareto set. The traditional method solves multi-objective optimization problems by comparing the fitness of solution in current population and recommending an optimal solution to users. However, when dealing with Web services composition problems, an optimal solution is difficult to meet the specific needs of users because of some fraudulent services and a variety of services. Therefore, the dissertation constructs Pareto set to solve this problem. The result is a non-dominated set, solving services composition problem preferably.Then, this dissertation improves the original algorithm. When selecting a solution, the original algorithm uses roulette strategy. It makes algorithm have premature convergence, which gets a poor diversity of population. Therefore, this dissertation adopts Bolzmann strategy as selection strategy to improve the original algorithm. It makes the algorithm have a better global search capability and improves the diversity of the population. Meanwhile, the nectar abandon strategy is adjusted accordingly and the dissertation modifies the search formula to accommodate services composition problems.At last, the dissertation designs a simulation experiment to verify the feasibility of the improved method. The result proves that the proposed method can effeciently avoid "Early Mature" phenomena, increasing the diversity of population. The experiment shows that the improved method can preferably solve Web services composition optimization issues.