PartisanLens: A Multilingual Dataset of Hyperpartisan and Conspiratorial Immigration Narratives in European Media

TL;DR

PartisanLens introduces the first multilingual dataset of European immigration headlines (1617 items across Spanish, Italian, and Portuguese) annotated for hyperpartisanship, PRCT, stance, and three rhetorical biases. The authors benchmark transformer and LLM-based approaches under zero-shot, few-shot, and supervised fine-tuning while also exploring LLMs as surrogate annotators and ideology-informed user-personas to mimic annotator diversity. Findings show PRCT signals are stronger and easier to detect than hyperpartisan cues, while stance remains the most challenging; supervision and larger models help, but LLMs still underperform human annotators, though persona conditioning can improve alignment for PRCT and, with stronger models, stance. The resource release and methodological insights offer a foundation for multilingual detection of partisan narratives in European contexts and highlight directions for extending language coverage, context depth, and cross-lingual transfer to improve robustness in misinformation detection.

Abstract

Detecting hyperpartisan narratives and Population Replacement Conspiracy Theories (PRCT) is essential to addressing the spread of misinformation. These complex narratives pose a significant threat, as hyperpartisanship drives political polarisation and institutional distrust, while PRCTs directly motivate real-world extremist violence, making their identification critical for social cohesion and public safety. However, existing resources are scarce, predominantly English-centric, and often analyse hyperpartisanship, stance, and rhetorical bias in isolation rather than as interrelated aspects of political discourse. To bridge this gap, we introduce \textsc{PartisanLens}, the first multilingual dataset of \num{1617} hyperpartisan news headlines in Spanish, Italian, and Portuguese, annotated in multiple political discourse aspects. We first evaluate the classification performance of widely used Large Language Models (LLMs) on this dataset, establishing robust baselines for the classification of hyperpartisan and PRCT narratives. In addition, we assess the viability of using LLMs as automatic annotators for this task, analysing their ability to approximate human annotation. Results highlight both their potential and current limitations. Next, moving beyond standard judgments, we explore whether LLMs can emulate human annotation patterns by conditioning them on socio-economic and ideological profiles that simulate annotator perspectives. At last, we provide our resources and evaluation, \textsc{PartisanLens} supports future research on detecting partisan and conspiratorial narratives in European contexts.

Figures (4)

• Figure 1: Hyperpartisan and PRCT stance proportions.
• Figure 2: Hyperpartisan and PRCT content by language.
• Figure 3: Example of a rationale-enhanced annotation.
• Figure 4: Example of the custom HTML platform used for the annotation process.