Filtering out improper accounts from Twitter user accounts for discovering individuals interested in certain topic

Abstract

Nowadays, the continuously growing popularity of Web-based Debates and online survey are requiring participants from different places and backgrounds which raises the need of the application for topic-oriented participants recommendation based on social networking service (SNS). Among several kinds of SNS, Twitter holds over 300 million users located all over the world. Therefore, the aim of this paper is to propose a filter technique applying natural language processing skill, which filters the individual users from other official or robot users of twitter for Web-based Debate system and online survey participants recommendation. To keep the balance of user groups for training, the search was divided into two parts with different list of keywords. As result, we gained the F measure of 0.93 by processing the data with SVM.