The explosive development of social networking tools like twitter has caused a tremendous influx of spam-like content and false accounts of users that can really interfere with the credibility of information and trust in the users. Malicious users and spam tweets have become a burning issue because of the sheer amount, speed and dynamism of data in social media.This project will suggest the Real Time Spam Filtering and Fake User Detection System employing the Extreme Machine Learning (ELM). First, common machine learning algorithms like the Random Forest and the Naive Bayes were applied and analyzed. Nevertheless, these models had a relatively lower accuracy and higher training time when using high pastoral and behavioral qualities.To overcome these shortcomings, it uses Extreme Machine Learning (ELM) which is a rapid learning algorithm of Single Hidden Layer Feedforward Neural Networks (SLFN). The proposed system is characterized by: data collection, preprocessing (removal of stop-words, tokenization, normalization), feature extraction (textual, profile and behavioral features extraction based on TF-IDF), and classification with ELM which leads to much less training time and better performance in generalization. It is experimentally established that the ELM-based model has higher accuracy, high training speed, and high scalability than the traditional classifiers.The presented framework offers an efficient, scalable and high-performance framework to improve real-time spam detection and fake user identification on social media sites.
SRAVANTI et al. (Sun,) studied this question.