Synthetic Multimodal Nudging Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/paxnea/LLM-multimodal-nudging
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了20名用户、40条消息以及50张精选图片及其对应标题，旨在为合成多模态引导任务构建个性化内容推荐。为确保数据多样性并避免偏见，该数据集采用了不同的大型语言模型进行生成。它适用于研究引导和超个性化领域。具体规模为：20名用户、40条消息、50张图片，任务内容为多模态内容的零样本推荐。
This dataset comprises 20 users, 40 messages, and 50 curated images paired with their respective captions, and is developed to construct personalized content recommendations for synthetic multimodal guidance tasks. To ensure data diversity and mitigate biases, the dataset was generated using multiple distinct large language models. It is suitable for research in the domains of guidance and hyper-personalization. Its specific specifications are as follows: 20 users, 40 messages, 50 images, and the task involves zero-shot recommendation of multimodal content.
提供机构：
Authors of the paper
