| 研究実績の概要 |
(1) This project uses big data data and computational methods to detect the expressed opinion on social media. We find the individuals’ tendency to maintain their partisan identification can lead to hostility toward individuals of opposing partisans, serving as a potential mechanism that contributes to polarization.
(2) This project explores the possibilities of LLMs (Large Language Models) in research related to opinion polarization. We have validated and demonstrated that LLMs can, through prompts, mimic individuals with specific partisan leanings and cognitive biases. This implies that LLMs can be used in social simulations to enrich the fidelity and complexity of simulations, potentially yielding deeper insights into the mechanisms of opinion polarization.
|
