{
  "id": 7402,
  "url": "https://arxiv.org/abs/2603.10351v1",
  "title": "Mitigating Translationese Bias in Multilingual LLM-as-a-Judge via Disentangled Information Bottleneck",
  "summary": "Large language models (LLMs) have become a standard for multilingual evaluation, yet they exhibit a severe systematic translationese bias. In this paper, translationese bias is characterized as LLMs systematically favoring machine-translated text over human-authored references, particularly in low-resource languages. We attribute this bias to spurious correlations with (i) latent manifold alignment with English and (ii) cross-lingual predictability. To mitigate this bias, we propose DIBJudge, a ",
  "authors": "Hongbin Zhang, Kehai Chen, Xuefen Bai, Youcheng Pan, Yang Xiang, Jinpeng Wang et al.",
  "category": "research",
  "topics": "bias-fairness,safety-alignment",
  "orgs": null,
  "regions": null,
  "published_at": "2026-03-11T02:55:29.000Z",
  "fetched_at": "2026-07-14T16:33:12.390Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/7402",
  "original_url": "https://arxiv.org/abs/2603.10351v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}