User security is the main aspect in present days from person to person interaction. In this paper we propose to introduce efficient user security mechanism in transmission of messages from one user to another user with secret message sharing. This is achieved through a flexible rule-based system, that allows users to customize the filtering criteria to be applied to their walls, and a Machine Learning based soft classifier automatically labeling messages in support of content-based filtering. In this process sending messages from one user to another in social networks there is a problem in accessing services in privately and publicly. In this process of serving user communication in social networks there is tremendous problem in profanity messages on user walls. In this paper we describe to develop methodology for security in their user walls with private messages. Our experimental results show efficient learning process in extracting feedback process generation in customized system operations in online social networks. In this we also consider the functionalities of each user in relevant data sharing of social networks.