SM bots have become ubiquitous on many platforms, particularly on Twitter, which is among the leading SM sources for traders and financial analysts. Though Twitter is adamant that bots represent less than 5% of all accounts, there is a far larger proportion of bot-generated posts on the platform. It is estimated that bots may account for
nearly 30% of all US-generated Twitter traffic.
Identifying traffic that is bot-generated vs human-generated is critical for noise reduction. PUMP uses a range of NLP-based techniques to dissect SM content into bot vs human-driven. Our system analyses the content produced by the originating account as well as the account’s network linking it to other SM accounts. In many cases, bots have a distinctly different network linkage structure compared to a human account.