In the specified documents data mining technique has been deprecated for filtering the OSN wall with unwanted messages or any type of vulgar messages. OSN is Online Social Network which has become an important part of the people life these days. People communicate over it with friends, relatives over a OSN wall. Thus to provide a feel of security to users personal stuff it is important to filter the OSN wall for any unwanted message .But the questions Arises, how to filter the OSN wall with a facility provided of blocking unwanted messages on the user's private wall. This can be gained through the flexible rule-based system which implements filtering criteria that can be customized by the user and a Machine Learning-based soft classifier which automatically labels messages in the support of content-based filtering .This paper consist of a literature survey paper of the existing system with proposed system as a technique to filter similar meaning words using Ontology along with the basic functionality to filter the OSN wall for unwanted message. In this paper a technique to build a social network with filtered message is elaborated. General TermsOn-line Social Network, short text classifier.

In text documents data mining techniques have been discovered for mining useful patterns. But there are some questions, how to properly use and update discovered patterns is still an open research issue, specifically in the text mining. Therefore most existing text mining methods have used term-based approaches but they all suffer from the problems of polysemy (multiple meaning word) and synonymy(same meaning word).This is the literature survey paper with proposed system develops innovative and successful patternbased technique which contains the processes of pattern taxonomy, pattern deploying and gradually developing pattern ,to improve the effectiveness of using and updating researched patterns for finding applicable and interesting data with effectual patterns as per the users requirements. In this paper user is also getting the meaningful information without wrong meaning problem.