International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169 Volume: 2 Issue: 11 3511 - 3513 _______________________________________________________________________________________________ 3511 IJRITCC | November 2014, Available @ http://www.ijritcc.org _______________________________________________________________________________________ Survey on a Rule Based System to Refine User Walls Tirgul Aniket Nandkumar Department of Computer Engineering, DGOI,FOE,Daund. Pune Maharashtra India anikettirgul333@gmail.com Salunke Shrikant Dadasaheb Department of Computer Engineering, DGOI,FOE,Daund. Pune Maharashtra India shrikantsalunke25@gmail.com Rajpure Amol Subhash Department of Computer Engineering, DGOI,FOE,Daund. Pune Maharashtra India amolrajpure9@gmail.com Bere Sachin Sukhadeo Department of Computer Engineering, DGOI,FOE,Daund. Pune Maharashtra India sachinbere@gmail.com Abstract -The core problem in today’s Online Social Networks (OSNs) is to allocate users the authority to manage the messages posted on their private space to avert that unwanted content. The unwanted data may contain political, vulgar, non-neural etc. message filtering systems are designed for unstructured or semi-structured data, as opposed to database applications, which use very structured data. In this paper, we proposed a System with the flexible rules to filter the unwanted messages posted on user wall. After passing threshold value, the informing message is sent to that user. This allows users to customize the refining criteria to be applied to their walls, and a Machine Learning-based classifier automatically classifies the messages and labelling messages in support of content-based filtering. Keywords - flexible rules, message filtering, online social networks, short text classification. __________________________________________________*****_________________________________________________ I. INTRODUCTION Online Social Networks (OSNs) square measure now a day’s one amongst the most popular interactive medium to speak, share, and disperse a substantial quantity of human life information. Daily and continuous communications imply the exchange of many sorts of content, as well as free text, image, audio, and video information. Consistent with facebookstatistics1 average user creates ninety items of content every month, whereas over thirty billion items of content (web links, news stories, blog posts, notes, icon albums, etc.) are parceled monthly. The massive and dynamic character of this information creates the premise for the utilization of internet content mining ways aimed to mechanically discover useful info dormant inside the info. They are instrumental to produce a lively support in advanced and sophisticated tasks concerned in OSN management, such as for instance access management or info filtering. Information filtering has been greatly explored for what considerations textual documents and, additional recently, website (e.g., [1], [2], [3]). However, the aim of the bulk of those proposals is especially to supply users a classification mechanism to avoid they're powerless by useless information. In OSNs, data filtering may be used for a different, additional sensitive, purpose. This can be thanks to the very fact that, in OSNs, there’s the likelihood of posting or commenting alternative posts on specific public/private areas, called normally walls. Data filtering will so be used to provide users the flexibility to mechanically management the messages written on their own walls, by filtering out unwanted messages. We have a tendency to believe that this can be a key OSN service that has not been provided thus far. Indeed, today OSNs offer little support to forestall unwanted messages on users walls. For instance, Facebook permits users to state United Nations agency is allowed to insert messages in their walls (i.e., friends, friends of friends, or outlined teams of friends). However, no content-based preferences area unit supported, and therefore, it’s insufferable to forestall sought messages, like political or vulgar ones, regardless of the user United Nations agency posts them. Providing this service isn't solely a matter of victimization antecedently outlined website mining techniques for a special application, rather it needs to design impromptu classification ways. This can be as a result of wall messages area unit official by short text that ancient classification strategies have serious limitations since short texts don't offer sufficient word occurrences. The aim of this work is so to propose and experimentally value an automatic system, referred to as Filtered Wall (FW), able to filter unwanted messages from OSN user walls. We have a tendency to exploit Machine Learning (ML) text categorization techniques [4] to mechanically assign with every short text message a group of classes supported its content. The major efforts in building a strong short text classifier (STC) area unit targeted within the extraction and choice of a set of characterizing and discriminant options. The solutions investigated during this paper area unit A Nextension of these adopted in an exceedingly previous work by US [5] from that we have a tendency to inherit the learning model, and