New benchmark dataset DECODEM enables automated extraction of corporate governance data

Researchers have introduced DECODEM, a new benchmark dataset designed to evaluate the automated extraction of corporate governance variables from organizational documents. The paper details experiments using large language models (LLMs) on this dataset, showing that automated extraction is feasible with high accuracy for many provisions. While frontier models perform well, the study also indicates that pipeline design can help bridge the capability gap between advanced and efficiency-focused models. AI

IMPACT This benchmark could accelerate the use of LLMs in legal research and corporate governance analysis.

RANK_REASON The cluster contains a research paper introducing a new benchmark dataset and evaluating LLM performance on it. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• Jens Frankenreiter

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →