From Heart to Words: Balancing Form and Function in the Expression of Empathy

ACL ARR 2024 December Submission1250 Authors

Abstract: Empathy is crucial for emotionally connecting with others and providing support, a need that has grown in online communities. While empathy involves understanding others' feelings, effectively communicating that understanding is equally important. In this study, we propose a novel approach to empathetic response generation by combining figurative language with manually annotated empathy causes to address both the linguistic form and semantic context. By integrating these elements, our approach improves multiple dimensions of empathetic responses, achieving a 7.6% improvement in BLEU, a 36.7% reduction in Perplexity, and a 7.6% increase in lexical diversity (D-1 and D-2) in automated evaluations compared to models without these features. Additionally, human assessments show a 24.2% increase in empathy ratings over the same baseline. These findings highlight the synergy between figurative language and empathy causes, offering valuable insights for enhancing empathetic communication across domains.

Paper Type: Long

Research Area: NLP Applications

Research Area Keywords: empathetic response generation, figurative language, cause-aware generation

Contribution Types: Model analysis & interpretability, NLP engineering experiment, Data resources, Data analysis

Languages Studied: Python

Submission Number: 1250