PLAtE

PLAtE是由弗吉尼亚大学和亚马逊的研究团队创建的大规模数据集，专注于购物领域的多商品评论页面信息提取。该数据集包含52,898个商品，来自6,694个页面，总计156,014个属性，旨在解决神经模型在半结构化网站信息提取中的训练数据不足问题。数据集通过多阶段方法收集和标注，适用于评估和训练网页提取模型，特别是在产品列表分割和属性提取任务上。

PLAtE is a large-scale dataset developed by research teams from the University of Virginia and Amazon, focusing on information extraction from multi-product review pages in the e-commerce shopping domain. This dataset contains 52,898 products sourced from 6,694 web pages, with a total of 156,014 attributes, and aims to address the shortage of training data for neural models in information extraction from semi-structured websites. The dataset is collected and annotated via a multi-stage workflow, and is applicable for evaluating and training web information extraction models, especially for product list segmentation and attribute extraction tasks.