Machine Learning Techniques to Evaluate Whether Twitter Accounts Are Human or Robot

Abstract

As social media becomes more relevant in daily life; efforts have been made to try to harness this relevancy and use it for nefarious goals. Bot accounts have been created by third-party organizations to try to influence public opinion, impersonate humans, and perform other forms of exploitation. Researchers have been concentrating efforts to attempt to recognize these accounts and label them so that the public can differentiate between genuine accounts and bot accounts. Our contributions to this research include the study of various classification algorithms on our dataset to identify the best performing algorithm for this sort of data. We also studied the inclusion of tweet time gap variance to attempt to capture potential unnatural patterns or consistency in time between tweets. Our experimental evaluation shows the feasibility of the proposed approach.