In this paper,we focus on understanding and modeling users retweeting behavior through implicit social relationships and its effects on information diffusion on Twitter social network.To achieve this goal,we first define two types of influences between users as Authority degree and Familiarity degree,corresponding to the unidirectional-follow and bidirectional-follow relationship on Twitter social network.And then we consider the influence of a tweet from its retweeted number and retweeting similarity.The combination of user influence and twitter influence together determine which tweet to be retweeted through the interactions between users.Further,we set up a tweets queue with limited size to model a users limited attention,because one can only pay attention to a portion of tweets shared on his/her homepage.On this basis,we build an agent-based model which simulates the users retweeting behavior-based information propagation by incorporating both user influence and tweet influence into the key process of selecting a tweet.By comparing statistical characteristics of multiplex networks between the simulated result and actual data,we have confirmed the applicability of our proposed model and methods on exploring retweeting behaviors through implicit social relationships on Twitter.