An Unsophisticated Neural Bots and Gender Profiling System

In recent years a sharp increase of bot-aided campaigns can be observed across social media networks. As a consequence, an own research discipline known as social bot detection has been established, to counteract these. In the context of the shared task "Bots and Gender Profiling" at the PAN workshop, we propose a simple neural network-based approach that determines for a given Twitter feed whether its author is a bot or a human, where in the latter case it distinguishes between male and female authors. On the official English test set, our approach achieves an accuracy of 92% and 83% for type and gender detection, respectively. For the Spanish test set, however, the results are lower (82% for type and 74% for gender detection).