Abstract
The abuse of web bots poses a great threat to daily life. There are lots of methods proposed to detect web bots. However, these web detection methods focus on specific application tasks such as chat bot, game bot, spam bot, and so on. In this paper, a web bot detection model based on mouse dynamics is proposed. Mouse dynamics, which analyzes user’s behavioral patterns, has been proven very effective in distinguishing human users from web bots. We propose a new time series representation method that combines position differences and directional speed values in one time step simultaneously to cover the raw mouse movement sequence into suitable input formats for deep learning models. Experimental results demonstrate that our method outperforms existing machine learning methods with handcrafted features and the deep learning method with visualization representation with a detection accuracy of 99.78% for the bot.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call