ClimRetrieve

ClimRetrieve是由苏黎世大学等机构创建的用于评估企业气候披露信息检索系统的基准数据集。该数据集包含超过8.5K个独特的问答对，模拟了可持续发展分析师的工作流程，旨在解决气候变化领域的信息检索问题。数据集通过专家标注，涵盖了从30份可持续发展报告中提取的详细气候相关问题及其答案，适用于评估和改进检索增强生成（RAG）系统，特别是在知识密集型的气候变化通信领域。ClimRetrieve is a benchmark dataset developed by institutions including the University of Zurich for evaluating enterprise climate disclosure information retrieval systems. This dataset contains over 8.5K unique question-answer pairs, simulating the workflow of sustainability analysts and targeting information retrieval tasks in the climate change domain. Expert-annotated, it covers detailed climate-related questions and their corresponding answers extracted from 30 sustainability reports. The dataset is applicable for evaluating and enhancing retrieval-augmented generation (RAG) systems, especially in knowledge-intensive climate change communication scenarios.