New dataset Strategic 16K benchmarks AI models for document sensitivity

Researchers have developed Strategic 16K, a new dataset of 16,000 diplomatic cables from WikiLeaks, designed to prevent label leakage in document sensitivity classification. This corpus was used to benchmark classical and transformer-based models, revealing that BERT achieved the highest accuracy (89.14%) and F1 score (89.33%) on the cleaned data. While transformer models like BERT and ELECTRA performed best, TF-IDF with logistic regression offered a strong performance at a lower computational cost. AI

IMPACT Establishes a new benchmark for document sensitivity classification, highlighting the importance of clean data for reliable AI performance.

RANK_REASON The cluster contains an academic paper detailing a new dataset and benchmarking results for AI models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• BERT
• ELECTRA
• Hugging Face
• logistic regression model
• Strategic 16K
• tf–idf
• WikiLeaks Public Library of US Diplomacy (PlusD)

• paper
• model release

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