BotBuster: Multi-platform Bot Detection Using A Mixture of Experts

Understand

Despite rapid development, current bot detection models still face challenges in dealing with incomplete data and cross-platform applications.

• In this paper, we propose BotBuster, a social bot detector built with the concept of a mixture of experts approach.
• Each expert is trained to analyze a portion of account information, e.g.
• username, and are combined to estimate the probability that the account is a bot.