Abstract: Focusing on Web service selection problem， a new modified ant colony optimization (ACO) algorithm is proposed. Both a nonlinear dynamic parameter of the pseudorandom proportion selection rule and a multipleoptimalsolution randomweighted route selection algorithm are employed in the algorithm proposed to control the behavior of ant colony. Besides， a fivedimensional service quality vector and the fitness function are used in the algorithm to evaluate the ant solutions， and each ant updates the pheromone according to the quality of their solutions they built. With these measures， the evolution ability of ant colony can be significantly improved. The experimental results show that the proposed algorithm outperforms traditional ACO algorithms.

Key words: service selection, ant colony optimization, randomweighted route selection, dynamic pseudorandom proportion selection parameter, algorithm performance evaluation index