Jd.com Chinese fake review dataset

The data set is divided into three parts. The first part is the review text crawled by the team from Jd.com comments, including more than 30,000 pieces of review data from mobile phones, clothing and other fields on Jd.com platform from 2014 to 2021. About 10,000 pieces of valid comment data were finally obtained after eliminating incorrect text and garbled characters. In the second part, in order to verify the effectiveness of the model for cross-domain experiments, the data extracted again is the review data of lj.com in the four fields of mobile phone, food, home appliances and clothing in 2023, among which, each field contains 200 reviews, and the ratio of real reviews to fake reviews is approximately 1:1. In the third part, two Chinese data sets (harassing short messages and microblog rumors) are used to verify the validity of the model for Chinese text classification.

本数据集共分为三个部分。第一部分为本团队从京东（Jd.com）平台评论区爬取的评论文本，涵盖2014年至2021年京东平台手机、服饰等多个领域的超3万条原始评论数据，经剔除无效文本与乱码后，最终得到约1万条有效评论数据。第二部分为验证模型在跨域实验中的有效性，本次额外提取了2023年lj.com平台手机、食品、家电、服饰四大领域的评论文本，每个领域包含200条评论，且真实评论与虚假评论的比例约为1:1。第三部分采用两个中文数据集（骚扰短信与微博谣言），用于验证模型在中文文本分类任务中的有效性。

搜集汇总

背景与挑战

背景概述

该数据集包含京东平台多个领域的真实与虚假评论数据，以及用于跨领域实验和中文文本分类验证的辅助数据集，适用于虚假评论检测和中文文本分类研究。