50个类别，每个类别50个样本，用于文本分类的数据集，解决多类单标签问题-50 categories are each 50 samples, used for text classification data sets to address the problem of multi-class single-label