In this paper, we present a bot detection algorithm to identify Twitter bot accounts and to classify them as positive, negative or neutral bots. It’s no surprise that bots are ubiquitous in social media. However, they can be problematic if they intentionally aim to manipulate valid information and spread misinformation, which can negatively affect users’ opinion of various topics. The complexity of the problem lies in the fact that bots are becoming increasingly similar to humans, in order to avoid detection. We present an unsupervised lexicon based approach to determine the polarity of tweets.
The main objectives of this paper are to verify Botometer’s ability to detect bots from a small set of Twitter accounts with accuracy. In addition to using a lexicon based approach to analyze the sentiment of tweets.