Twitter Spam Detection In Text Using Hybrid A...

Twitter Spam Detection In Text Using Hybrid Approach

Author Name : Shalini Ranjan , Supriya P. Panda

ABSTRACT

People today have grown accustomed to the nonstop flow of information. Popular social networking sites like Twitter, Facebook, and Quora are facing a serious issue due to the popularity of spam accounts on these sites. These accounts are made to mislead unsuspecting genuine users into clicking on risky links or continually sending redundant messages by using bots. As a result, spam detection is introduced. An LSTM-based model for spam identification is proposed in this work, and it will deliver pertinent tweets based on the user's prior experience. Several models, including Gaussian Naive Bayes, Logistic Regression (LR), K-Nearest Neighbor (KNN), and LSTM, are compared in this study. According to the data, LSTM outperforms 90.6percent other text-based algorithms including Gaussian Naive Bayes, Logistic Regression, and KNN, which have accuracy rates of 87.2 percent, 86.6 percent, and 86.5 percent, respectively.

Keywords: K-Nearest Neighbor, Long Short Term Memory, Logistic Regression, Spam Detection.