ChronoQA

<b>ChronoQA: A Benchmark for Temporal-Sensitive Question Answering</b><br>ChronoQA is a novel dataset designed to evaluate and advance retrieval-augmented generation (RAG) systems in handling time-sensitive queries. Featuring diverse question types—including explicit/implicit time expressions, multi-hop reasoning, and varying temporal scopes—ChronoQA is automatically constructed from dynamic news sources to ensure scalability and up-to-date relevance. With over 5,000 high-quality QA pairs, it supports single- and multi-document reasoning, addressing critical gaps in existing benchmarks by emphasizing temporal alignment and complex event dependencies. ChronoQA serves as a foundational resource for developing robust models capable of navigating evolving real-world knowledge.<br>Paper Title: Chrono-O1: Advancing Temporal Sensitive RAG through Progressive Multi-Step Reflection<b>ChronoQA：面向时序敏感型问答的基准数据集</b><br>ChronoQA是一款全新的基准数据集，旨在评估并优化检索增强生成（Retrieval-Augmented Generation, RAG）系统处理时序敏感查询的能力。该数据集涵盖多样化的问答类型，包括显式/隐式时间表述、多步推理以及不同的时序范围；其自动从动态新闻源构建而成，以保障可扩展性与实时相关性。该数据集包含超过5000条高质量问答对，支持单文档与多文档推理，并通过强调时序对齐与复杂事件依赖关系，弥补了现有基准数据集的关键短板。ChronoQA可作为开发能够适配不断演进的真实世界知识的稳健模型的基础资源。<br>论文标题：Chrono-O1：通过渐进式多步反思优化时序敏感型RAG系统