Metraipolis › archive › dc_f352d9d1dbd2

MUCnoHARM@GermEval Shared Task 2026: Retrieval-based In-Context Learning for Defamatory Offences, and Where It Falls Short — arXiv:2609.09791

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.09791
License (as archived)
CC0-1.0
Archived
2026-10-16T07:53:21.265Z
SHA-256 of the archived text
1fcfd576f776220087a5ec4f13666e588f0a80776f7b64d541dc67eaa1d47ab4
Stable URL of this page
https://metraipolis.com/documento/dc_f352d9d1dbd2

How to cite

MUCnoHARM@GermEval Shared Task 2026: Retrieval-based In-Context Learning for Defamatory Offences, and Where It Falls Short — arXiv:2609.09791 — arXiv (en), https://arxiv.org/abs/2609.09791 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f352d9d1dbd2 (Metraipolis archive dc_f352d9d1dbd2, sha256 1fcfd576f776220087a5ec4f13666e588f0a80776f7b64d541dc67eaa1d47ab4)
BibTeX
@misc{dcf352d9d1dbd2,
  title = {MUCnoHARM@GermEval Shared Task 2026: Retrieval-based In-Context Learning for Defamatory Offences, and Where It Falls Short — arXiv:2609.09791},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.09791},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f352d9d1dbd2; sha256 1fcfd576f776220087a5ec4f13666e588f0a80776f7b64d541dc67eaa1d47ab4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.09791. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

With hate speech being ubiquitous online, automatic detection is crucial, in particular when it comes to criminally relevant social media posts. We study a variety of retrieval-based in-context learning (RetICL) strategies for detecting defamatory offences under §§ 185-187 StGB (the subject of GermEval 2026 Subtask 4). Few-shot prompting beats zero-shot, but retrieval-based approaches offer only marginal gains over random demonstrations, and even fall behind an optimised static set of demonstrations. Providing concrete legal knowledge helps, yet model choice outweighs every other system choice. Models over-predict criminal relevance while still missing 26-57% of criminally relevant posts, suiting them for triage rather than autonomous moderation.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-10 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.09791 — 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 2026-10-16T07:24:00.900Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-10?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).