Due to the increasing number of wireless mobile devices,the possibility of mobile communications without infrastructure becomes a reality.The Decentralized Mobile Social Network(DMSN) is a paradigm where nodes can move freely and organize themselves arbitrarily.Routing in these environments is difficult for the reason of the rapid changes of the social relationship graph’s topology.Meanwhile,the social ties among nodes change overtime.Therefore,an efficient data forwarding mechanism should be considered over the temporal weighted relationship graph.In this paper,an Advanced routing Protocol based on Parameters Optimization in the Weighted mobile social network(APPOW) is proposed to improve the delivery success ratio and reduce the cost of exchanging information.APPOW combines the normalized relative weights of three local social metrics,i.e.,LinkRank,similarity and contact strength,to select the next relay node.The weights of the three metrics are derived by pair-wise learning algorithm.The result shows that APPOW outperforms the state-ofthe-art SimBet Routing in delivering message and significantly reduces the average hops.Additionally,the delivery performance of APPOW is close to Epidemic Routing but without message duplications. Due to the increasing number of wireless mobile devices, the possibility of mobile communications without infrastructure becomes a reality. The Decentralized Mobile Social Network (DMSN) is a paradigm where nodes can move freely and organize themselves arbitrarily. Routing in these environments is difficult for the reason of the rapid changes of the social relationship graph’s topology.Meanwhile, the social ties among nodes change overtime. Beforefore, an efficient data forwarding mechanism should be considered over the temporal relationship pattern. In this paper, an Advanced routing Protocol based on Parameters Optimization in the Weighted mobile social network (APPOW) is proposed to improve the delivery success ratio and reduce the cost of exchanging information. APPOW combines the normalized relative weights of three local social metrics, ie, LinkRank, similarity and contact strength, to select the next relay node. The weights of the three metrics are derived by pair-wise learning algorith the result shows that APPOW outperforms the state-of-the-art SimBet Routing in delivering message and significantly reduces the average hops. Additionally, the delivery performance of APPOW is close to Epidemic Routing but without message duplications.