From Argumentation to Deliberation: Perspectivized Stance Vectors for Fine-grained (Dis)agreement Analysis

TL;DR

This paper introduces Perspectivized Stance Vectors (PSVs), a fine-grained representation that encodes issue-specific perspectives in arguments by mapping them to signature concepts from ConceptNet. It defines PSV construction, multiple stance-value prediction models (Baseline, RoBERTa, GPT4o), and aggregation schemes to derive perspectivized acceptability scores that distinguish agreement, orthogonality, and disagreement. Through extensive annotations on deliberative data and a large-scale unannotated evaluation, the approach demonstrates that PSVs can reveal per-concept agreement patterns and facilitate deliberation by identifying actionable anchors for consensus. The work emphasizes interpretability and controllability for deliberative decision making, while acknowledging data limitations and suggesting future directions toward end-to-end models and broader topic coverage.

Abstract

Debating over conflicting issues is a necessary first step towards resolving conflicts. However, intrinsic perspectives of an arguer are difficult to overcome by persuasive argumentation skills. Proceeding from a debate to a deliberative process, where we can identify actionable options for resolving a conflict requires a deeper analysis of arguments and the perspectives they are grounded in - as it is only from there that one can derive mutually agreeable resolution steps. In this work we develop a framework for a deliberative analysis of arguments in a computational argumentation setup. We conduct a fine-grained analysis of perspectivized stances expressed in the arguments of different arguers or stakeholders on a given issue, aiming not only to identify their opposing views, but also shared perspectives arising from their attitudes, values or needs. We formalize this analysis in Perspectivized Stance Vectors that characterize the individual perspectivized stances of all arguers on a given issue. We construct these vectors by determining issue- and argument-specific concepts, and predict an arguer's stance relative to each of them. The vectors allow us to measure a modulated (dis)agreement between arguers, structured by perspectives, which allows us to identify actionable points for conflict resolution, as a first step towards deliberation.

Figures (12)

• Figure 1: Example PSVs for 'Animal Hunting'.
• Figure 2: Disagreement ($\mathcal{P}_0$) distribution of argument pairs from the same stance or different stance.
• Figure 3: Acceptability scores computed with $\mathcal{P}_0$.
• Figure 4: Agreement scores among stakeholder groups for 'Animal Hunting'. Fig. \ref{['app:fig:cs:stakeholder_global']} shows results for other topics.
• Figure 5: (Dis)agreement of selected perspectives.