CITE

CITE数据集是由罗格斯大学创建的一个用于研究图像与文本间话语关系的资源。该数据集包含516个文档，主要来源于RecipeQA数据集，专注于烹饪食谱领域。数据集的创建过程涉及众包标注，通过特定的标注方案确保了标签的可靠性。CITE数据集的应用领域包括理解和生成多模态文档，旨在解决图像与文本间如何协同贡献于有效沟通的问题。

The CITE dataset is a resource developed by Rutgers University for researching discourse relationships between images and text. It comprises 516 documents primarily sourced from the RecipeQA dataset, with a focus on the culinary recipe domain. The development of this dataset involved crowdsourced annotation, and a specific annotation scheme was adopted to ensure the reliability of the labels. The application scope of the CITE dataset includes multimodal document understanding and generation, aiming to solve the problem of how images and text can synergistically contribute to effective communication.