Computer Science

Affect Recognition in Conversations Using Large Language Models

S. Feng, G. Sun, et al.

This study probes whether large language models can recognise human affect—emotions, moods, and feelings—in both open-domain chit‑chat and task‑oriented dialogues using IEMOCAP, EmoWOZ, and DAIC‑WOZ, evaluating zero‑shot and few‑shot in‑context learning as well as task‑specific fine‑tuning and the impact of ASR errors. The research was conducted by the authors: Shutong Feng, Guangzhi Sun, Nurul Lubis, Wen Wu, Chao Zhang, and Milica Gašić.