Rethinking stance detection: A theoretically-informed research agenda for user-level inference using language models

TL;DR

This paper critiques current stance detection approaches, emphasizing the need for a theoretical framework and user-level analysis, and explores how large language models can enhance inference of individual attributes.

Contribution

It introduces a research agenda that integrates theoretical perspectives and user-level data into stance detection using large language models.

Findings

Large language models can infer user-level stance attributes.

Current stance detection lacks theoretical grounding and user-level focus.

A four-point agenda guides future research for more inclusive and impactful stance detection.

Abstract

Stance detection has emerged as a popular task in natural language processing research, enabled largely by the abundance of target-specific social media data. While there has been considerable research on the development of stance detection models, datasets, and application, we highlight important gaps pertaining to (i) a lack of theoretical conceptualization of stance, and (ii) the treatment of stance at an individual- or user-level, as opposed to message-level. In this paper, we first review the interdisciplinary origins of stance as an individual-level construct to highlight relevant attributes (e.g., psychological features) that might be useful to incorporate in stance detection models. Further, we argue that recent pre-trained and large language models (LLMs) might offer a way to flexibly infer such user-level attributes and/or incorporate them in modelling stance. To better illustrate this, we briefly review and synthesize the emerging corpus of studies on using LLMs for inferring stance, and specifically on incorporating user attributes in such tasks. We conclude by proposing a four-point agenda for pursuing stance detection research that is theoretically informed, inclusive, and practically impactful.