Metraipolis › archive › dc_f0603fdecdec

Learning Sexism Detection Using Multi-Agent Perspectivist Preference Optimization — arXiv:2608.04056

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.04056
License (as archived): CC0-1.0
Archived: 2026-10-31T11:24:32.542Z
SHA-256 of the archived text: 398b96b2379b90af388ae6e71c96231f985f6213cdc902331ae76a7f3b2cd806
Stable URL of this page: https://metraipolis.com/documento/dc_f0603fdecdec

How to cite

Learning Sexism Detection Using Multi-Agent Perspectivist Preference Optimization — arXiv:2608.04056 — arXiv (en), https://arxiv.org/abs/2608.04056 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f0603fdecdec (Metraipolis archive dc_f0603fdecdec, sha256 398b96b2379b90af388ae6e71c96231f985f6213cdc902331ae76a7f3b2cd806)

BibTeX

@misc{dcf0603fdecdec,
  title = {Learning Sexism Detection Using Multi-Agent Perspectivist Preference Optimization — arXiv:2608.04056},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.04056},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f0603fdecdec; sha256 398b96b2379b90af388ae6e71c96231f985f6213cdc902331ae76a7f3b2cd806},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.04056. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

When people label text for sexism, they often disagree, and not because some of them are wrong: they genuinely perceive sexism differently. Most NLP systems discard this disagreement by collapsing it into a majority vote. We propose the Multi-Agent Perspectivist Preference Optimization (MAP-PO) framework to keep these different perspectives. On the EXIST 2024 dataset of labeled English and Spanish tweets, we first cluster annotators by their labeling behavior rather than their demographic attributes. We then fine-tune one Large Language Model agent per cluster to reproduce that cluster's annotation behavior, and coordinate the agents with preference optimization that combines individual and team-level rewards. We evaluate MAP-PO in four settings defined by two languages and two backbone language models, asking whether each agent reproduces the annotations of its own cluster and whether the agents together reproduce the majority label. Two findings hold in all four settings. First, without fine-tuning the agents behave almost identically, so cluster-specific training is necessary. Second, we show that training each agent only on the labels of its own cluster pushes the agents far beyond the clusters they should represent, while adding a shared team-level training signal consistently keeps each agent calibrated to its cluster.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-06 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Computers and Society (cs.CY); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.04056 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-31T11:13:23.152Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-06?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).