TSN4PI Datasets
收藏资源简介:
TSN4PI Datasets是由香港科技大学(广州)、中国科学技术大学和香港科技大学联合构建的两个大规模政治意识形态数据集,旨在解决社交媒体政治话语分析中的数据稀缺问题。其中,新闻数据集包含约23万篇来自AllSides的新闻文章,覆盖2017至2022年,并附有连续意识形态评分;社交媒体数据集包含来自X(Twitter)平台约7700万条帖子,涉及4545名用户,时间跨度从2007至2022年。数据集通过爬取和弱监督标注方式创建,新闻文章继承来源的意识形态标签,社交媒体帖子则通过用户互动和内容分析进行标注。该数据集主要用于训练和评估政治意识形态检测与演化预测模型,助力研究政治极化、回声室效应及意识形态动态,为非商业研究提供公开资源。
TSN4PI Datasets are two large-scale political ideology datasets jointly constructed by The Hong Kong University of Science and Technology (Guangzhou), University of Science and Technology of China, and The Hong Kong University of Science and Technology, aiming to address the data scarcity issue in political discourse analysis on social media. The news dataset contains approximately 230,000 news articles from AllSides, spanning from 2017 to 2022, with continuous ideological scores attached. The social media dataset includes around 77 million posts from the X (formerly Twitter) platform, involving 4545 users, with a time span from 2007 to 2022. The datasets were created through web crawling and weakly-supervised annotation: news articles inherit the ideological tags from their original source, while social media posts are annotated via user interaction and content analysis. These datasets are primarily used for training and evaluating models for political ideology detection and evolution prediction, facilitating research on political polarization, echo chamber effects, and ideological dynamics, and serving as an open resource for non-commercial research.
- 1Against Political Polarization: A Unified Framework for Tracing Evolving Political Ideologies on Social Media香港科技大学(广州); 中国科学技术大学; 香港科技大学 · 2026年



