ECKGBench
收藏资源简介:
ECKGBench是一个专门为评估大型语言模型在电商领域知识上的能力而设计的基准数据集。该数据集基于大规模电商知识图谱构建,包含大量真实的三元组(实体-关系-实体),旨在通过自动生成的问题来评估LLM模型的准确性。数据集的问题生成和负样本采样均采用自动化流程,同时融合了电商领域的专业知识,确保了评估的质量和效率。ECKGBench可应用于评估LLM模型在电商领域的知识边界,推动基础模型的开发和评估。
ECKGBench is a benchmark dataset specifically designed to evaluate the capabilities of large language models (LLMs) on e-commerce domain knowledge. Built on a large-scale e-commerce knowledge graph, this dataset contains a vast number of real triples (entity-relation-entity). It aims to assess the accuracy of LLM models via automatically generated questions. Both the question generation and negative sample sampling of the dataset adopt automated workflows, while integrating professional knowledge from the e-commerce domain to ensure the quality and efficiency of the evaluation. ECKGBench can be applied to evaluate the knowledge boundaries of LLM models in the e-commerce domain, promoting the development and evaluation of foundation models.




