ChronoQA: A Benchmark Dataset for Temporal Reasoning in RAG Systems

Summary

Researchers have introduced ChronoQA, a benchmark dataset designed to evaluate temporal reasoning capabilities in Retrieval-Augmented Generation (RAG) systems, published in Nature Scientific Data. The dataset focuses on Chinese question answering and addresses limitations in existing temporal QA benchmarks, which typically support only direct temporal logic, lack diversity in question types (such as aggregate or implicit time expressions), and rarely require multi-document reasoning. ChronoQA i