K-Act2Emo: Korean Commonsense Knowledge Graph for Indirect Emotional Expression
Kyuhee Kim, Surin Lee, Sangah Lee
TL;DR
This work tackles the challenge of inferring emotions from indirect cues in Korean literary and narrative contexts by building K-Act2Emo, a Korean Commonsense Knowledge Graph focused on indirect emotional expressions. It defines a three-way reasoning taxonomy (PosEnv, NegEnv, NonEmo) and collects 1,900 head expressions with 6,002 tails through a two-step crowdsourcing process, ensuring culturally grounded inferences. The paper compares K-Act2Emo with ATOMIC20_20, demonstrates rich coverage of emotional cues, and shows that fine-tuning COMET-BART on K-Act2Emo yields superior emotional-inference performance, approaching GPT-4 Turbo on automatic metrics and human judgments. This dataset and methodology advance emotion understanding in Korean NLP and offer a robust resource for improving language models' interpretation of indirect emotional narratives in culturally specific contexts.
Abstract
In many literary texts, emotions are indirectly conveyed through descriptions of actions, facial expressions, and appearances, necessitating emotion inference for narrative understanding. In this paper, we introduce K-Act2Emo, a Korean commonsense knowledge graph (CSKG) comprising 1,900 indirect emotional expressions and the emotions inferable from them. We categorize reasoning types into inferences in positive situations, inferences in negative situations, and inferences when expressions do not serve as emotional cues. Unlike existing CSKGs, K-Act2Emo specializes in emotional contexts, and experimental results validate its effectiveness for training emotion inference models. Significantly, the BART-based knowledge model fine-tuned with K-Act2Emo outperforms various existing Korean large language models, achieving performance levels comparable to GPT-4 Turbo.