Classification based credibility analysis on twitter data

Abstract

Twitter is an important place to get access to breaking news and information. So, it is necessary to check the trustworthiness of tweets. Credibility is used to assess the quality of being believable or worthy of trust. Credibility analysis refers to attempt to an ascertain truthfulness in short lie detection. In this work, credibility of the twitter data can be assessed using centrality measures. First the tweets are preprocessed using the preprocessing techniques. The preprocessing techniques on tweets: a stop word removal, stemming, pos tagging etc. are used to improve the performance. Preprocessed twitter data can be used to identify the tweet and author features. Then the centrality measures are applied to the preprocessed dataset. The proposed centrality measures used in this work are Betweeness centrality, Eigenvector centrality, Degree centrality and Closeness centrality. The centrality measures are used to find out the trust between the users and it will be given as input to classifiers. The classifiers like Naïve Bayesian, Support Vector Machine and K Nearest Neighbor are used to classify the tweets based on credibility.