In this paper, we try to use linguistic reputation system based on Computing with Words instead of a fix number to rate the Peer in Peer-to-Peer network since the intuition of human. For meeting the security requirement of P2P, we utilize Secret Sharing scheme to store the reputations and consider the weight of price of every tradeoff to obtain the last global reputation of one seller. At the end of paper, we give the experiment to show how the price factor effect the last global reputation. We also give the Time Cost of this new linguistic reputation.