EmoLASP: Emotion Recognition with Language Models and Answer Set Programming
Artificial IntelligenceComputation and Language
Summary
The authors present EmoLASP, a new method that combines language models with a type of logic reasoning called Answer Set Programming to better guess emotions in conversations. They tested it on a common emotion dataset and found it works better than just using the language models alone, especially when the models don’t get to see the full conversation history. This approach helps reduce the need to fine-tune or feed long dialogue histories into the models. However, once a model is fine-tuned and sees the conversation context, the reasoning part contributes less.
Authors
Thao Le, Michael Thielscher
Abstract
Emotion recognition in conversations is increasingly tackled with language models, but these models can be unstable and expensive to fine-tune or to prompt with long dialogue histories. We propose EmoLASP, a framework that combines a language model with declarative reasoning via Answer Set Programming (ASP) to predict VAD scores (Valence-Arousal-Dominance) in conversations. Experiments on a widely used benchmark dataset (IEMOCAP) across six open-source LLMs (3B-120B) and two PLMs (BERT, RoBERTa) show that EmoLASP improves prediction performance compared to using the language model alone, even when the LLMs/PLMs are given no dialogue history in their prompts or input vectors. The gains are largest for prompt-only LLMs, which EmoLASP uses without any fine-tuning. However, for fine-tuned PLMs, the reasoner adds little once dialogue history is available. EmoLASP's LLM pipeline demonstrates the potential advantages of using a reasoning approach to ensure emotion prediction consistency and to reduce both the cost of fine-tuning and the cost of prompting with long dialogue histories.