Metraipolis › archive › dc_f42129cc6233

From Abusive Language Classification to Sequence Labeling Identification — arXiv:2610.06287

Source
arXiv
Original
https://arxiv.org/abs/2610.06287
License (as archived)
CC0-1.0
Archived
2026-10-06T05:28:31.381Z
SHA-256 of the archived text
13f343b1fa17399edb9a1278d10326170619a20ed2411a56890842c5d4f243b9
Stable URL of this page
https://metraipolis.com/documento/dc_f42129cc6233

How to cite

From Abusive Language Classification to Sequence Labeling Identification — arXiv:2610.06287 — arXiv, https://arxiv.org/abs/2610.06287 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f42129cc6233 (Metraipolis archive dc_f42129cc6233, sha256 13f343b1fa17399edb9a1278d10326170619a20ed2411a56890842c5d4f243b9)
BibTeX
@misc{dcf42129cc6233,
  title = {From Abusive Language Classification to Sequence Labeling Identification — arXiv:2610.06287},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.06287},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f42129cc6233; sha256 13f343b1fa17399edb9a1278d10326170619a20ed2411a56890842c5d4f243b9},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.06287. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Industrial content moderation must process massive message streams under tight latency constraints, yet most abusive language (AL) detection systems rely on sentence-level classification (ALC), which neither localizes abusive spans nor identifies who is targeted. We define Abusive Language Identification (ALI) as a sequence-labeling task that jointly extracts AL spans and target mentions, and assess whether this approach can be used for text moderation. On a pilot corpus drawn from a production moderation pipeline, we compare ALI with ALC on cross-domain generalization and implicit abuse, and we also evaluate AL and target span detection. ALI remains competitive with ALC while providing localized outputs for moderators, with a modest and configuration-sensitive advantage on implicit abuse. Exact AL boundaries and target spans remain difficult to recover. We complement this comparison with a qualitative analysis and discuss perspectives on complete target--span linking and on structured benchmarks for ALI.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.06287 — 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 2026-10-06T05:24:20.366Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).