yuqiangJEP/jep-papers-and-corpus

JEP论文和评估语料库是一个包含JEP/HJS/JAC/COE问责框架研究论文和评估材料的数据集。研究论文部分包括5篇不同定位的论文，涵盖数学基础层、协议架构层、记录隐私层、因果链层和认知共识层。评估语料库部分包含24个场景单元（92个事件点），用于表达充分性测试，覆盖了多种应用场景，如代理辅助研究、多代理任务执行、代理介导的客户支持、基于代理的采购和人类监督的自主操作，并设定了编码目标，包括覆盖率>90%、残差其他<10%和编码者间一致性（Cohens kappa）≥0.75。

The JEP Papers and Evaluation Corpus is a dataset containing research papers and evaluation materials for the JEP/HJS/JAC/COE accountability framework. The papers section includes 5 papers with different focuses, covering the mathematical foundation layer, protocol architecture layer (JEP), record privacy layer (predecessor), causal chain layer, and cognitive consensus layer. The evaluation corpus consists of 24 scenario units (92 event points) for expressive adequacy testing, covering various application scenarios such as agent-assisted research, multi-agent task execution, agent-mediated customer support, agent-based procurement, and human-supervised autonomous operations. Coding targets include a coverage rate > 90%, residual other < 10%, and intercoder agreement (Cohens kappa) ≥ 0.75.