Abstract: Notary mechanism is based on the cross-chain mechanism of credit endorsement node. Aimed at the single credit evaluation problem of notary endorsement node, it is proposed to apply the weighted LeaderRank algorithm to the evaluation model. The algorithm calculated the node credit weight value by collecting the historical transaction evaluation JP+1information of nodes, participated in the trust ordering algorithm, got the safe and trustworthy notary node, and made the notary mechanism more stable and trustworthy. The results show that the improved weighted LeaderRank algorithm makes comprehensive use of node historical transaction evaluation information and transaction trust relationship, which is of great significance to the accurate selection of notary nodes and the maintenance of the safe and reliable notary mechanism.