In this generation, using online social network (OSN) is an unavoidable powerful weapon to exhibit peoples' views and ideas. The users depending upon their interests can select the persons who must post/comment messages in their wall. The present excavation in OSN user wall is "No filtering of abusive messages". That is the selected persons can post any sort of messages in their wall. So in this paper, we propose a filtered wall to permeate offensive messages using rule based and text classification techniques. We have evaluated the performance using metrics, from which it is shown that proposed method is better. © 2015 The Authors. Published by Elsevier B.V.