Abstract
Many studies have been proposed to discover the deceptive information on different social networking sites, especially on Twitter, and in many languages. For the disclosure of credibility in Arabic publications, few works have been published. In this paper, we propose a smart classifier that determines the credibility of Arabic "tweets" posted on Twitter. This classifier integrates socialmining and natural language processing techniques. Several (user- and content-related) features have been used in the training and testing phases for the proposed model. We have added the ability to check for tweets by comparing them to Google search results. We have also developed an algorithm to check the similarity of the user's screen and his/her official name on Twitter. By combining these new features with the most common features of credibility detection, our proposed classifier identifies the credibility of the tweets in the dataset we are utilizing and outperforms similar works in the field associated with the detection of the credibility of Arabic tweets in terms of accuracy.