The Rashomon Wikipedia: A Data-Perspectivist Analysis of Divergent Historical Narratives

2026-09-27

AI Summary

This study addresses how historical narrative biases and citation segregation in multilingual Wikipedia fragment controversial events into divided epistemic communities. By analyzing Wikipedia articles across five languages, this work integrates human annotation, large language models (LLMs), citation stance classification, and longitudinal time-series analysis to quantify narrative evolution and reveal geopolitical influences. The core contribution is the proposal of "Peacemaker," an automated reconciliation framework that leverages adversarial prompting to mitigate model hallucinations. Empirical findings demonstrate significant citation segregation; for instance, only 2 out of 119 citations are shared across language editions regarding the Battle of Posada. Furthermore, this research validates the framework's effectiveness in generating neutral consensus while preserving the attribution of divergent perspectives, offering a scalable approach to bridging epistemic divides in collaborative knowledge platforms.

Abstract

Wikipedia aims to provide a unified, neutral record of history, yet its independent language editions often function as distinct epistemic communities, creating divergent narratives around contested events. This paper investigates cross-lingual historiographical bias by analyzing Wikipedia articles across five languages (Romanian, Hungarian, Russian, Turkish, and English) focusing on three contentious events in Romanian history: the Battle of Posada (1330), the Soviet occupation of Bessarabia (1940), and the Night Attack at Targoviste (1462). Using human annotators and Large Language Models (LLMs) to classify citation stance and quantify narrative evolution from 2005 to 2024, we identify a phenomenon of "citation isolation". In the case of the Battle of Posada, only 2 out of 119 citations were shared between language editions, with the Romanian edition exhibiting a 91% pro-national bias compared to the balanced Hungarian edition. Longitudinal analysis reveals that these narratives are volatile and responsive to contemporary geopolitics, evidenced by a significant shift in the Russian framing of Bessarabia in 2024. Finally, we propose a "Peace-Maker" pipeline to automate conflict reconciliation. We demonstrate that while standard prompting leads models to hallucinate consensus, "adversarial" prompting, which explicitly instructs the model to preserve and attribute disagreement, achieves near-perfect neutrality scores.

Problem

• cross-lingual historiographical bias
• Wikipedia
• divergent narratives
• citation isolation
• contested events

Innovation

• Cross-lingual bias
• Large Language Models
• Citation isolation
• Adversarial prompting
• Conflict reconciliation