Public Data

该数据集包含了170万条创意信息，来自50万个商品项目。由于标题数据不可用，我们采用了每个项目的图像特征平均值来模拟标题信息。数据规模为170万条创意和50万个商品项目，任务是对点击率进行预测。

This dataset contains 1.7 million pieces of creative information derived from 500,000 product items. As the title data was unavailable, we adopted the average image features of each item to simulate the title information. With a scale of 1.7 million creative entries and 500,000 product items, the core task of this dataset is click-through rate (CTR) prediction.