five

Pitt Ads dataset

收藏
arXiv2025-09-30 收录
下载链接:
https://people.cs.pitt.edu/~kovashka/ads/
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64,832条广告,每条广告都由3位不同的专家标注者用英语撰写了3条解释。这些解释旨在捕捉广告背后的说服信息。此外,每条广告都与15条信息配对,其中包括与该广告对应的3条正面信息,以及从不同广告标注中随机抽取的12条负面信息。该数据集的规模为64,832条广告,每条广告配有3条解释,适用于广告理解图像到文本检索任务。

This dataset contains 64,832 advertisements. Each advertisement is accompanied by three explanations written in English by three distinct expert annotators, designed to capture the persuasive messaging behind the advertisements. Additionally, each advertisement is paired with 15 pieces of information, including three positive messages corresponding to the target advertisement and 12 negative messages randomly sampled from annotations of other advertisements. This dataset, consisting of 64,832 advertisements each paired with three explanatory texts, is applicable to advertisement understanding image-to-text retrieval tasks.
提供机构:
University of Pittsburgh
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作