ABSTRACT / The strategy that social bots use to interact with people in online social networks significantly influences the mesoscopic structures of the networks they form. This talk will provide details about an experimental approach we employed, seeking to observe network motifs in bot-human interactions. We employed three different strategies: Trend-targeting (TTS), user-targeting (UTS) and keyword-targeting (KTS). We observed significant differences both in the successfulness and in the network motif statistics as a function of the strategies. While there was an overlap in the motifs generated with KTS and TTS, it was only the UTS which resulted in more complex motifs. The most rewarded strategy (TTS) exhibited the least diverse local structure patterns, whereas, conversely, the least-rewarded strategy (UTS) had the largest set of local structures, characterized by as many as six different types of motifs. This study contributes to the understanding of the human-bot ecosystem and to future design of efficient bots.
BIO / Abdullah is a PhD candidate in network science at the Central European University and holds a master’s degree in engineering. He is primarily interested in the real-world use of Data Analytics, including quantification, text mining, and natural language processing. Currently, he is researching and testing social and political bots and how they interact with humans as part of an ecosystem. He is also exploring computational propaganda, the impact of language models, and digital extremism networks.