Words Like Knives: Backstory-Personalized Modeling and Detection of Violent Communication

2025-05-27

AI Summary

This study addresses the neglect of relational context and affective nuance in NLP models for detecting intimate partner violence (IPV) communication, proposing a relation-aware conflict identification paradigm. We introduce PersonaConflicts, a novel corpus of 5,772 simulated dialogues spanning familial and romantic conflict scenarios. It is the first to systematically incorporate relational backstories and to adopt fine-grained, turn-level annotations grounded in Nonviolent Communication (NVC) theory, alongside a background modulation effect evaluation framework. Through human–model comparative experiments, LLM reasoning analysis, and multi-dimensional affective–relational assessment, we find that relational background polarity significantly influences human judgment—yet current LLMs fail to effectively leverage such information and consistently overestimate the positivity of utterances. These findings expose a critical limitation of LLMs in real-world relationship mediation tasks.

Technology Category

• Natural Language Processing: Discourse, Pragmatics & Argument Mining
• Cognitive Modeling & Cognitive Systems: Affective Computing
• Application Domains: Humanities & Computational Social Science

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search

Abstract

Conversational breakdowns in close relationships are deeply shaped by personal histories and emotional context, yet most NLP research treats conflict detection as a general task, overlooking the relational dynamics that influence how messages are perceived. In this work, we leverage nonviolent communication (NVC) theory to evaluate LLMs in detecting conversational breakdowns and assessing how relationship backstory influences both human and model perception of conflicts. Given the sensitivity and scarcity of real-world datasets featuring conflict between familiar social partners with rich personal backstories, we contribute the PersonaConflicts Corpus, a dataset of N=5,772 naturalistic simulated dialogues spanning diverse conflict scenarios between friends, family members, and romantic partners. Through a controlled human study, we annotate a subset of dialogues and obtain fine-grained labels of communication breakdown types on individual turns, and assess the impact of backstory on human and model perception of conflict in conversation. We find that the polarity of relationship backstories significantly shifted human perception of communication breakdowns and impressions of the social partners, yet models struggle to meaningfully leverage those backstories in the detection task. Additionally, we find that models consistently overestimate how positively a message will make a listener feel. Our findings underscore the critical role of personalization to relationship contexts in enabling LLMs to serve as effective mediators in human communication for authentic connection.

Problem

• Detecting violent communication in close relationships considering personal histories
• Evaluating LLMs in conflict detection using nonviolent communication theory
• Assessing impact of relationship backstory on human and model perception

Innovation

• Leveraging NVC theory for conflict detection
• Introducing PersonaConflicts Corpus dataset
• Assessing backstory impact on perception