The online social networks (OSN) offers proficient message controls that are posted on their private space in order to avoid un-preferred content displayed to users. But, OSN provides a low supportive and flexibility to the private message in the user’s own space. Most of the existing works on OSN presents a system to facilitate user by discarding the messages posted on their private wall. The existing works also develops flexible rule based system with the filtering criteria to post only useful messages on their walls for user satisfaction. The machine learning based classifier presents involuntarily labeling of messages that support content based filtering. Moreover, OSN filters the message content based on message originator relationship and characteristics. OSN offer the rule layer with classification module with the semantics for filtering policy to improve the domain.
The drawback of existing work is that filters did not concern about the user privacy demand. The provisions for the feedbacks on the online learning as well as the nomenclature of the users in the black list are not mentioned. Hence, the challenge lies in establishing better filtering technique. The proposed work is user privacy concerned social proximity rules to develop an efficient filter for open social network. Proximity rule estimation (PRE) provides private and verifiable proximity computation based on polynomial secure share. PRE address the user privacy concern by getting the feedback forms from the user for friend-like nomenclature. Proposed PRE is able to discover the potential friends with the preferred message exchange between them using relational privacy index. And also helps to identify a variety of possible attack messages by analyzing real traces. Additionally, PRE develops novel solution for secure proximity estimation and allows user to identify the acceptable messages that are listed in their social network space. Proposed work intends in fulfilling user privacy needs and demands in designing OSN filters. The adequate numbers are given for the privacy feedbacks on the online learning and the user nomenclature are exclusively inserted in the black list for classification of filter rules. Experimental evaluations are conducted to prove the better performance of PRE in terms of attack density, minimum execution time and less memory space utilized.