Harnessing the Power of Multi-Classifiers: A Novel Adaptive Feature Neural Network Framework for Twitter Spam Detection

Abstract

Currently, there are several social media platforms such as Facebook, Twitter, and Instagram on the internet that bring people together. Twitter, known for its wealth of information, is a popular choice among users for connecting with others and sharing updates. The platform utilizes Google Safe-browsing to identify and block spam URLs. The advanced API of Twitter allows for easy data manipulation, attracting various spammers to the platform. Previous studies have implemented ML (Machine Learning) techniques to combat Twitter spam, but their algorithms lack comprehensive evaluation and accuracy when dealing with large datasets. To address these challenges, this study proposes an adaptive feature neural network analysis. The proposed model is developed by most efficient method which differentiates between spam and non-spam tweets. In proposed model, the classifier is applied to a large dataset of 600 million public tweets and evaluated based on various metrics such as accuracy, True Positive Rate (TPR), False Positive Rate (FPR), and F-measure. Results indicate that the proposed technique shows robust performance in spam detection on Twitter.