文档敏感性分类基准：经典与Transformer模型对比研究
原标题：Benchmarking Classical and Transformer-Based Models for Document Sensitivity Classification
AI 摘要
本研究针对组织文档敏感性分类中的标签泄漏问题，构建了包含16,000份外交电报的Strategic 16K基准数据集，并系统评估了六种模型架构。在清洁基准上，BERT表现最佳（准确率89.14%，F1 89.33%），而TF-IDF结合逻辑回归在较低计算成本下达到最强经典模型性能。该工作首次在明确泄漏控制条件下提供了可复现的敏感性分类基准。
正文节选
Benchmarking Classical and Transformer-Based Models for Document Sensitivity Classification Abstract Automatic sensitivity classification of organizational documents is a critical yet underserved problem, where the consequences of misclassification range from regulatory violations to security breaches. While AI-based approaches offer a scalable alternative to manual review, their reliability depends fundamentally on the integrity of training data. A pervasive but underreported problem in this do