ChronoQA
收藏资源简介:
ChronoQA 是一个大规模且系统构建的数据集,专为评估时间敏感的检索增强生成(RAG)系统而设计。该数据集由 2019 年至 2024 年间发布的超过 300,000 篇新闻文章构建而成,包含 5,176 个问题,涵盖绝对、聚合和相对时间类型,具有显式和隐式时间表达式。数据集支持单文档和多文档场景,反映了现实世界对时间对齐和逻辑一致性的要求。ChronoQA 通过广泛的时态任务提供结构化评估,为在不断发展中的知识基准测试 RAG 系统提供了动态、可靠和可扩展的资源。
ChronoQA is a large-scale, systematically constructed dataset specifically designed for evaluating time-sensitive retrieval-augmented generation (RAG) systems. It is constructed from over 300,000 news articles published between 2019 and 2024, and contains 5,176 questions covering absolute, aggregated, and relative temporal types with both explicit and implicit temporal expressions. The dataset supports both single-document and multi-document scenarios, reflecting real-world requirements for temporal alignment and logical consistency. ChronoQA provides structured evaluation via a wide range of temporal tasks, offering a dynamic, reliable, and scalable resource for benchmarking RAG systems against evolving world knowledge.




