Spot the Bot: Identify Texts Generated by Bots

The rise in social media popularity has also coupled with widespread fraud using bot accounts that distribute computer-generated content, misleading users into thinking about a brand or popular political views. In this thesis, particular interest is the linguistic analysis of such texts to develop a new method for detecting bots. We propose a new method that uses the texts’ representation in the form of a semantic space, where each word corresponds to a vector of the occurrence frequency of this word in a particular set of texts. The analysis of semantic spaces will be done using the clustering algorithm. We will test the hypothesis that clustering metrics, such as compactness and distance between clusters, will differ between people and bots, which will become the basis for solving the problem of content classification.