FairytaleQA

FairytaleQA是一个专注于幼儿园至八年级学生叙事理解的数据集。由教育专家基于证据理论框架生成，包含10,580个明确和隐含问题，源自278个儿童友好的故事，涵盖七种叙事元素或关系。该数据集不仅有助于评估模型的细粒度学习技能，还支持教育领域的问题生成任务。通过与问题生成模型的基准测试，显示该模型在FairytaleQA上训练后能提出高质量和更多样化的问题。

FairytaleQA is a dataset focused on narrative comprehension for students ranging from kindergarten to eighth grade. Developed by educational experts based on an evidence-based theoretical framework, it contains 10,580 explicit and implicit questions derived from 278 child-friendly stories covering seven narrative elements or relationships. This dataset not only helps evaluate the fine-grained learning skills of models, but also supports question generation tasks in the education field. Benchmark tests on question generation models show that models trained on FairytaleQA can generate high-quality and more diverse questions.