Answers are grounded in this paper and cite their sources.
In recent years, the computing community has made significant efforts to develop socially intelligent machines capable of understanding social context in ways similar to humans. However, comparatively little attention has been devoted to developing approaches that can automatically identify conflict or an individual's conflict management style (CMS), defined as a person's attitude and behavior in response to conflict. The five primary CMSs identified in the literature are avoiding, dominating, obliging, compromising, and integrating. Traditional methods for assessing these styles typically rely on self-report questionnaires or observer-based evaluations, both of which are susceptible to bias and inaccuracies. This article presents a machine learning approach for inferring an individual's CMS (CMS) from typing behavior, that is, from the way people type during interactive text-based conversations. Typing behavior may provide a more objective measure of a person's preferred approach to conflict resolution because it reflects unconscious behavioral patterns. We evaluate the proposed approach on a corpus of 30 dyadic chat sessions involving 60 previously unacquainted participants, comprising a total of 191 375 keystrokes. The results demonstrate that typing behavior is an effective predictor of CMS, achieving F1 scores of up to 93%, depending on the specific style being inferred.
No takes yet. Share an insight, caveat, or question.
Buker et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: