Using RL to Identify Divisive Perspectives Improves LLMs Abilities to Identify Communities on Social Media
This work addresses the challenge of adapting LLMs for social media analysis, but it is incremental as it focuses on prompting enhancements rather than fundamental breakthroughs.
The paper tackled the problem of identifying user communities on social media by using a smaller LLM to improve prompting of larger black-box LLMs, resulting in experimental improvements on Reddit and Twitter for tasks like community detection, bot detection, and news media profiling.
The large scale usage of social media, combined with its significant impact, has made it increasingly important to understand it. In particular, identifying user communities, can be helpful for many downstream tasks. However, particularly when models are trained on past data and tested on future, doing this is difficult. In this paper, we hypothesize to take advantage of Large Language Models (LLMs), to better identify user communities. Due to the fact that many LLMs, such as ChatGPT, are fixed and must be treated as black-boxes, we propose an approach to better prompt them, by training a smaller LLM to do this. We devise strategies to train this smaller model, showing how it can improve the larger LLMs ability to detect communities. Experimental results show improvements on Reddit and Twitter data, on the tasks of community detection, bot detection, and news media profiling.