SciEGQA：一个用于科学证据基础问题回答与推理的数据集
SciEGQA: A Dataset for Scientific Evidence-Grounded Question Answering and Reasoning
AI总结：
SciEGQA数据集通过精细标注和大规模自动构建，提升视觉语言模型在科学文档中的证据定位与推理能力。
摘要：
科学文档包含复杂的多模态结构，这使得在文档视觉问答中进行证据定位和科学推理特别具有挑战性。然而，大多数现有基准仅在页面级别评估模型，而没有显式标注支持答案的证据区域，这限制了可解释性和评估的可靠性。为了解决这一限制，我们引入了SciEGQA，一个具有语义证据定位的科学文档问题回答与推理数据集，其中支持证据被表示为语义连贯的文档区域，并通过边界框进行标注。SciEGQA包含两个组成部分：一个包含1,623对高质量问题-答案对的人工标注细粒度基准，以及一个通过自动化数据构建流水线生成超过30,000对问答对的大规模自动构建训练集。对广泛视觉语言模型（VLMs）的大量实验表明，现有模型在科学文档中的证据定位和基于证据的问题回答上仍存在困难。在所提出数据集上训练显著提高了VLMs的科学推理能力。项目页面可在https://yuwenhan07.github.io/SciEGQA-project/上访问。
英文摘要：
Scientific documents contain complex multimodal structures, which makes evidence localization and scientific reasoning in Document Visual Question Answering particularly challenging. However, most existing benchmarks evaluate models only at the page level without explicitly annotating the evidence regions that support the answer, which limits both interpretability and the reliability of evaluation. To address this limitation, we introduce SciEGQA, a scientific document question answering and reasoning dataset with semantic evidence grounding, where supporting evidence is represented as semantically coherent document regions annotated with bounding boxes. SciEGQA consists of two components: a **human-annotated fine-grained benchmark** containing 1,623 high-quality question--answer pairs, and a **large-scale automatically constructed training set** with over 30K QA pairs generated through an automated data construction pipeline. Extensive experiments on a wide range of Vision-Language Models (VLMs) show that existing models still struggle with evidence localization and evidence-based question answering in scientific documents. Training on the proposed dataset significantly improves the scientific reasoning capabilities of VLMs. The project page is available at https://yuwenhan07.github.io/SciEGQA-project/.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
