CONVERSATIONAL PAPERS (cPAPERS)
收藏资源简介:
CONVERSATIONAL PAPERS (cPAPERS) 是一个包含对话式问答对的数据集,专注于科学论文中的图表、方程和表格。该数据集由佐治亚理工学院AI虚拟助理实验室创建,包含5030对问答,源自OpenReview上的评审和反驳。数据集的创建过程涉及从OpenReview收集问答对,并将其与arXiv上的科学文章相关联。cPAPERS旨在通过提供多模态、情境化的对话数据,支持开发能够理解和回应科学文档中复杂内容的对话助手。
CONVERSATIONAL PAPERS (cPAPERS) is a dataset of conversational question-answer pairs focused on figures, equations and tables within scientific papers. Developed by the AI Virtual Assistant Lab at the Georgia Institute of Technology, the dataset contains 5030 question-answer pairs derived from reviews and rebuttals hosted on OpenReview. The dataset construction workflow entails collecting question-answer pairs from OpenReview and associating them with corresponding scientific articles on arXiv. cPAPERS aims to support the development of conversational assistants capable of comprehending and responding to complex content in scientific documents by providing multimodal, contextual conversational data.

- 1cPAPERS: A Dataset of Situated and Multimodal Interactive Conversations in Scientific Papers佐治亚理工学院AI虚拟助理实验室 · 2024年



