TwiUSD: A Benchmark Dataset and Structure-Aware LLM Framework for User Stance Detection

TL;DR

This paper introduces TwiUSD, a large-scale benchmark for user stance detection that combines content and social structure, and proposes MRFG, a structure-aware framework leveraging LLMs and graph neural networks to improve detection accuracy.

Contribution

The paper presents TwiUSD, the first comprehensive UserSD benchmark with social links, and introduces MRFG, a novel structure-aware framework utilizing LLM relevance filtering and adaptive feature routing.

Findings

MRFG outperforms baseline models in stance detection tasks.

TwiUSD provides high-quality, large-scale data for UserSD research.

MRFG effectively handles noise and heterogeneity in social network data.

Abstract

User-level stance detection (UserSD) remains challenging due to the lack of high-quality benchmarks that jointly capture linguistic and social structure. In this paper, we introduce TwiUSD, the first large-scale, manually annotated UserSD benchmark with explicit followee relationships, containing 16,211 users and 47,757 tweets. TwiUSD enables rigorous evaluation of stance models by integrating tweet content and social links, with superior scale and annotation quality. Building on this resource, we propose MRFG: a structure-aware framework that uses LLM-based relevance filtering and feature routing to address noise and context heterogeneity. MRFG employs multi-scale filtering and adaptively routes features through graph neural networks or multi-layer perceptrons based on topological informativeness. Experiments show MRFG consistently outperforms strong baselines (including PLMs, graph-based models, and LLM prompting) in both in-target and cross-target evaluation.