Abstract: OSN often encounters the issue of users’ control over their private messages. OSN does not completely support this requirement. This dissertation deals with the same issue, where a system is proposed, in which users directly can control the posted messages on their wall. For achieving this, a rule based system is designed which allows a user to manage their filtering criteria. Also the use of machine learning based soft classifier helps in automatically labeling the messages that supports content based filtering.

Keywords: On-line Social Networks, Machine learning, Information Filtering, Short Text Classification.