Abstract
The partner selection is an important decision problem in the formation of dynamic collaboration among Cloud Provides (CPs) To acquire optimal collaboration, both Individual and collaborative performance of candidate partners should be considered In the existing methods for partner selection, the collaborative performance is evaluated by using linear functions which cannot address the comprehensive relationships among candidate partners This paper proposes an evaluation approach using Back Propagation Neuron Network (BPNN) instead of any fixed objective function Through training, the BPNN can achieve function approximation and provide estimates of collaborative performance The experiment results show that our approach is effective and accurate in the evaluation.