Automatic Detection of Bot-generated Tweets

Abstract

Deep neural networks have the capacity to generate textual content which is increasingly difficult to distinguish from that produced by humans. Such content can be used in disinformation campaigns and its detrimental effects are amplified if it spreads on social networks. Here, we study the automatic detection of bot-generated Twitter messages. This task is difficult due to combination between the strong performance of recent deep language models and the limited length of tweets. In this study, we propose a challenging definition of the problem by making no assumption regarding the bot account, its network or the method used to generate the text. We devise two approaches for bot detection based on pretrained language models and create a new dataset of generated tweets to improve the performance of our classifier on recent text generation algorithms. The obtained results show that the generalization capabilities of the proposed classifier heavily depends on the dataset used to trained the model. Interestingly, the two automatic dataset augmentation proposed here show promising results. Their introduction leads to consistent performance gains compared to the use of the original dataset alone. CCS CONCEPTS• Applied computing → Document management and text processing.

Anatomy of an AI-powered malicious social botnet

“…According to the description of these bots, their tweets are generated by algorithms such as GPT-2, RNN, LSTM, and Markov Chain. Subsequent studies have built upon this dataset to explore various strategies for bot detection based on content (Saravani et al, 2021;Gambini et al, 2022;Tourille et al, 2022). Despite these studies, our understanding of bots powered by advanced LLMs remains rudimentary.…”

A Survey on the Detection and Impacts of Deepfakes in Visual, Audio, and Textual Formats

“…On the one hand, bots have positive uses, such as automating customer service responses, providing personal assistance, or automating routine posts for businesses on social media. However, the primary misuse of bots in textbased deepfakes involves the propagation of misinformation, spamming, or manipulation of public sentiment on social platforms [94], [95]. For instance, malicious social bots can be utilised during elections to amplify specific narratives or disseminate false information about candidates, thereby influencing public opinion [96].…”

MAD '22 Workshop: Multimedia AI against Disinformation

“…Cross-entropy loss is used to maximize the inter-class variations, while the center-loss is used to minimize the intra-class variations. Tourille et al [10] studies the automatic detection of bot-generated Twitter messages, which is in particular a difficult task due to the combination of the strong performance of recent deep language models and the limited length of tweets. They propose a definition of the task, devise two approaches based on pretrained language models, as well as the introduction of a new dataset of generated tweets.…”