Published Paper Details:

With increase in the popularity on social media and millions and billions of people using it every day. This popularity of applications like Facebook, Twitter and Instagram grabs the attention of spammers. As through these they can trap genuine users with malicious activities. There is a significant amount of research done in this field. The primary focus of these researches is generally based on the accounts or users whose activity poses them as suspicious. These activities include posting of the same content, posting tweets that have no relevance to the trending topics and tagging them as one of the trending topics, sending bulk direct messages or users that have similar contents and are created on the same day. However, much of the research done focuses on spam accounts. There is little to none research done based on a model that marks tweets as spam and along with a sentiment analysis. In the proposed system, we propose a Machine learning system that would detect tweets as spam or ham. This spam detection would be done considering factors such as: shortened URLs, Emails that lead to malicious sites etc. The tweets would also be marked as spam based on the language. Using NLP, we would form a system that would mark tweets as spam if they have the potential to hurt sentiments of other users. The model would be trained and tested on a previously labelled dataset. This model would then be incorporated in a website that would take tweets as an input from the user. The result would be creation of a model that would give the tweet as spam or not based on the sentiment and spammer tactics.