Extracting and Analyzing Factors to Identify the Malicious Conversational AI Bots on Twitter
Gitika Vyas, Prathamesh Muzumdar, Anitha Chennamaneni, Anand Rajavat, Romil Rawat · 2024
On social media, many third-party vendors utilize Conversational Artificial Intelligence (CAI) to use social bots to spread marketing campaigns and to increase followers, thereby opening the door for malicious bots to compromise the security on social media, especially on Twitter where the posts are concise, making it hard to distinguish between human written posts and bot-generated tweets. Thus, such malicious bots on Twitter can break into user accounts, spread misinformation, breach account data, and market advertising spam. Hence, this chapter aims to conduct a detailed exploratory study to identify the crucial Twitter account features that help to detect malicious bots. Machine learning-based techniques such as information gain, correlation, and chi-square feature selections are used in this chapter to select the top feature set by comparing all three techniques. Twitter account data provided by Kaggle.com, which are publicly available, have been used. The finding suggests that Twitter account age, tweet replies, number of user mentions, friends count, favorites counts, listed count, account verification status, followers count, number of users who liked a tweet, default profile, bot description, and retweet count are the factors that can help to identify and detect the malicious bots on social media.