Service computing is a new paradigm and has been widely used in many fields. The multi-objective service selection is a basic problem in service computing and it is non-deterministic polynomial(NP)-hard. This paper pr...Service computing is a new paradigm and has been widely used in many fields. The multi-objective service selection is a basic problem in service computing and it is non-deterministic polynomial(NP)-hard. This paper proposes a novel multi-objective artificial bees colony(n-MOABC) algorithm to solve service selection problem. A composite service instance is a food source in the algorithm. The fitness of a food source is related to the quality of service(QoS) attributes of a composite service instance. The search strategy of the bees are based on dominance. If a food source has not been updated in successive maximum trial(Max Trial) times, it will be abandoned. In experiment phase, a parallel approach is used based on map-reduce framework for n-MOABC algorithm. The performance of the algorithm has been tested on a variety of data sets. The computational results demonstrate the effectiveness of our approach in comparison to a novel bi-ant colony optimization(NBACO)algorithm and co-evolution algorithm.展开更多
基金the National Natural Science Foundation of China(Nos.61202085,61300019)the Ningxia Hui Autonomous Region Natural Science Foundation(No.NZ13265)the Special Fund for Fundamental Research of Central Universities of Northeastern University(Nos.N120804001,N120204003)
文摘Service computing is a new paradigm and has been widely used in many fields. The multi-objective service selection is a basic problem in service computing and it is non-deterministic polynomial(NP)-hard. This paper proposes a novel multi-objective artificial bees colony(n-MOABC) algorithm to solve service selection problem. A composite service instance is a food source in the algorithm. The fitness of a food source is related to the quality of service(QoS) attributes of a composite service instance. The search strategy of the bees are based on dominance. If a food source has not been updated in successive maximum trial(Max Trial) times, it will be abandoned. In experiment phase, a parallel approach is used based on map-reduce framework for n-MOABC algorithm. The performance of the algorithm has been tested on a variety of data sets. The computational results demonstrate the effectiveness of our approach in comparison to a novel bi-ant colony optimization(NBACO)algorithm and co-evolution algorithm.