Brazilian Political Demonstration Dataset

Description

The Brazilian Political Protest Dataset (Annotated Tweets) is a collection of 5,000 manually labeled tweets related to protests in Brazil on September 7, 2021, and subsequent demonstrations in the following days. The dataset captures public discourse on Twitter, including opinions, news, and media content shared by users supporting and opposing the protests.

To collect the dataset, we used a keyword-based approach, selecting terms that were trending in Brazil at the time. The 5,000 annotated tweets were manually labeled to support research in political discourse analysis, misinformation detection, and social media studies. Due to the location and context of the protests, most tweets are in Portuguese, with a small portion in English and Spanish.

More details about the dataset can be found in:

Few-shot Learning for Multi-modal Social Media Event FilteringJosé Nascimento, João P. Cardenuto, Jing Yang, and Anderson RochaPublished in the 2022 IEEE International Workshop on Information Forensics and Security (WIFS)

IEEE Explorer | arXiv

Usage and Applications

This dataset might be valuable for research in:

Political Discourse Analysis: Understanding how different political groups interact online.

Misinformation & Fact-Checking: Analyzing fake news and manipulated media in protests.

Social Media Engagement & Opinion Mining: Investigating sentiment and polarization.

Multimodal AI Research: Studying how text, images, and news links contribute to online discourse.

Media Content

Due to the terms of use from the social networks, we do not make publicly available the texts and images that were collected. However, we can provide some extra piece of media content by contacting the authors.

Funding

DéjàVu thematic project, São Paulo Research Foundation (grants 2017/12646-3,  2019/04053-8, 2020/02241-9 and 2020/02211-2)