Hi everyone! Together with my research group, I’ve been working on a project for the past few months, and I think it might interest you all.<\/p>\n
YSocial<\/a> is a digital twin of a social network platform that improves the simulation of dynamic social interactions by integrating large language model (LLM) agents. <\/p>\n You can design your own scenario with LLM agents and describe them using multiple features, such as their political leaning, age, personality traits, interests, and more. Agents will interact on a topic of discussion (e.g., politics<\/a>, sports) and according to a desired follower\/content recommender system. Additionally, you can even make them discuss news extracted in real-time by RSS feeds!<\/p>\n For now, we have observed that many real-world phenomena emerge (without us forcing them), e.g., influencers, viral content, long-tailed distributions for users’ connections, etc.<\/p>\n This is just a sneak-peak of all YSocial features, you can read more on the website!<\/p>\n YSocial is on Github<\/a>, open and free for everyone! Feel free to give us some feedback and contribute to the project. There is also a preprint available on ArXiv<\/a> and a website<\/a> with some pre-made scenarios you can test. <\/p>\n<\/div>\n submitted by \/u\/Internal-Newspaper91 <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
