DigestAI news desk
hot OpenAI Unveils GPT‑6 Astra: Record‑Breaking 3D Rendering, Loop‑Transformer Architecture new Xi pledges China-led BRICS AI open-source community and digital ecosystem hot Mistral AI Secures $3.5B Series D, Valuation Tops $24B, Emphasizes Enterprise Control new LLM-Anchored Paralinguistic Boost for Alzheimer's Detection hot OpenAI launches Agents API in public beta, exposing Codex infrastructure to developers new Firmus Adds OpenAI; NVIDIA, Blackstone Face Different Sides of AI Buildout hot Meta launches Muse, a personal AI agent for everyday tasks on iOS and Android new AI could reshape capitalism, but the path depends on policy and ownership hot OpenAI board member warns company is not on track to prevent catastrophic AI loss of control new New probability-wave framework links trader behavior to AGI architecture design hot OpenAI claims 10,000 agents solved Navier-Stokes Millennium Prize problem in 88 hours new OpenAI pledges to match Anthropic's embedded evaluator safety commitment hot OpenAI agents uploaded malicious RubyGems packages in May, researchers say new Language Models Can't Detect Their Own Training Data hot Cognition Releases SWE-2 Coding Model Trained on Kimi K3 new Intern-S2-397B: Hugging Face's New Multimodal Foundation Model
Research updated

CMNIE Benchmark Introduces Structured Extraction for Chinese Military News

CMNIE is a new benchmark that focuses on structured information extraction from Chinese military news. The dataset contains 13,000 manually annotated documents, covering seven event types, ten argument roles, seven entity types, and eight relation types. It extends previous military‑domain resources by jointly modeling triggers, arguments, entities, and relations under a unified schema.

1 source primary source

Key points

  • CMNIE dataset: 13k Chinese military news, 7 event types, 10 argument roles, 7 entity types, 8 relation types.
  • Zero‑shot LLMs identify semantic units but fail to match gold span boundaries, making relation extraction the hardest task.
  • Benchmark promotes research on schema adherence, exact span matching, and joint extraction in specialized Chinese news.

The authors evaluate a range of approaches on a shared test set, including supervised extraction models, zero‑shot large language models, and fine‑tuned LLM‑based methods. Results show that while zero‑shot LLMs can identify relevant semantic units, they struggle to match gold span boundaries exactly, and relation extraction remains the most difficult task. The benchmark is intended to drive research into schema adherence, exact span matching, and joint extraction in specialized Chinese news.

Read the original at arXiv cs.CL · by Yan Yu, Mengna Zhu, Zhenyu Song, Hao Yang, Haiwen Chen, Mao Wang primary source Open source ↗

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us.

Comments

via GitHub Discussions

More in Research

All →

Related stories