During the last couple of years there have been many researches on Online Social Networks (OSN). The common manner of representing an OSN is by a user-based graph, where the vertices are different OSN users, and the edges are different interactions between these users, such as friendships, information-sharing instances, and other connection types. The question of whether a certain user is willing to share its information to other users, known and less known, is a question that occupies several researches in aspects of information security, sharing habits and information-flow models for OSN. While many approaches take into consideration the OSN graph edges as sharing-probability factors, here we present a novel approach, that also combines the vertices as well-defined attributed entities, that contain several properties, in which we seek a certain level of credibility based on the user’s attributes, such as number of total friends, age of user account, etc. The edges in our model represent the connection-strength of two users, by taking into consideration the attributes that represent their connection, such as number of mutual friend, friendship duration, etc. and the model also recognizes resemblance factors, meaning the number of similar user attributes. This approach optimizes the evaluation of users’ information-sharing willingness by deriving it from these attributes, thus creating an accurate flow-control graph that prevents information leakage from users to unwanted entities, such as adversaries or spammers. The novelty of the model is mainly its choice of integrated factors for user credibility and connection credibility, making it very useful for different OSN flow-control decisions and security permissions.