NormGenesis: Multicultural Dialogue Generation via Exemplar-Guided Social Norm Modeling and Violation Recovery

2025-09-22

AI Summary

This study addresses the limited adaptability of multilingual (English, Chinese, Korean) dialogue systems to social norm violations in cross-cultural interactions. Methodologically, we propose a novel “Violation-to-Resolution” (V2R) dialogue paradigm, implemented via an example-guided iterative refinement framework that jointly aligns linguistic, affective, and cultural dimensions early in generation to meet societal expectations. We introduce a multi-level annotation scheme—covering norm compliance, intent, and emotional response—and release a high-quality dataset comprising 10,800 multilingual dialogue turns. Our contributions are threefold: (1) the first formalization of the V2R paradigm for fine-grained social norm modeling and dynamic violation remediation; (2) significant improvements in response naturalness, cross-cultural generalization, and ethical sensitivity across diverse scenarios; and (3) consistent outperformance over state-of-the-art baselines on all evaluated metrics.

Technology Category

• Natural Language Processing: Safety and Robustness
• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Social norms govern culturally appropriate behavior in communication, enabling dialogue systems to produce responses that are not only coherent but also socially acceptable. We present NormGenesis, a multicultural framework for generating and annotating socially grounded dialogues across English, Chinese, and Korean. To model the dynamics of social interaction beyond static norm classification, we propose a novel dialogue type, Violation-to-Resolution (V2R), which models the progression of conversations following norm violations through recognition and socially appropriate repair. To improve pragmatic consistency in underrepresented languages, we implement an exemplar-based iterative refinement early in the dialogue synthesis process. This design introduces alignment with linguistic, emotional, and sociocultural expectations before full dialogue generation begins. Using this framework, we construct a dataset of 10,800 multi-turn dialogues annotated at the turn level for norm adherence, speaker intent, and emotional response. Human and LLM-based evaluations demonstrate that NormGenesis significantly outperforms existing datasets in refinement quality, dialogue naturalness, and generalization performance. We show that models trained on our V2R-augmented data exhibit improved pragmatic competence in ethically sensitive contexts. Our work establishes a new benchmark for culturally adaptive dialogue modeling and provides a scalable methodology for norm-aware generation across linguistically and culturally diverse languages.

Problem

• Modeling multicultural social norms for culturally appropriate dialogue generation
• Addressing social norm violations through recognition and repair in conversations
• Improving pragmatic consistency in underrepresented languages during dialogue synthesis

Innovation

• Violation-to-Resolution dialogue type for norm violation recovery
• Exemplar-based iterative refinement for pragmatic consistency
• Multicultural framework generating socially grounded dialogues