Fabrício Benevenuto

The fragmentation of public data in Brazil reflects broader sociotechnical challenges in digital governance, institutional coordination, and knowledge organization. Heterogeneous standards and limited interoperability restrict collective knowledge production, weaken evidence-based policymaking, and reinforce data asymmetries. We introduce Brazil Data Commons, a semantic infrastructure designed to integrate heterogeneous Brazilian public datasets within a shared knowledge framework. By adopting globally recognized ontologies and interoperable data standards, Brazil Data Commons contributes to the development of digital public infrastructure and connects Brazilian data to the broader Data Commons ecosystem. Beyond technical integration, the platform supports new forms of cross-domain analysis, public transparency, and data accessibility. Through intuitive interfaces and flexible access mechanisms, Brazil Data Commons lowers barriers to data use and strengthens the capacity of researchers, policymakers, and civic actors to generate data-driven insights. We hope our system may help researchers to correlate ‘offline’ social-economic indicators with online digital traces. This opens new avenues for digital demography, allowing researchers to integrate traditional data with social media and web activity to better understand Brazil’s complex population dynamics.

Online social networks (OSNs) serve as environments for various forms of social interaction, ranging from influencers interacting with their fans to more peer-to-peer forms of community debate. Instant messaging platforms are an example of spaces used for more horizontal peer-to-peer conversations. During election cycles, these digital spaces often become exposed to political manipulation and propaganda based on social ties. Regardless of their original purpose, politics can co-opt the group discourse in a process of politicization. Although studies explored the politicization of non-political topics such as vaccines and political coordination in OSNs, the politicization of non-political spaces themselves remains overlooked. This study investigates the politicization of non-political public Telegram groups in the context of the 2022 Brazilian elections. Using a curated set of Brazilian Telegram groups, we extract the data via the Telethon API and apply a BERT model to detect political messages. Our findings indicate a 60% increase in political content, primarily occurring in groups with themes indirectly related to politics (e.g., News) or in groups with more general-purpose themes that foster open-ended discussions (e.g., Leisure). This political content is concentrated in a few users, generating more engagement and further political debate.

The 2016 United States presidential election was marked by the abuse of targeted advertising on Facebook. Concerned with the risk of the same kind of abuse to happen in the 2018 Brazilian elections, we designed and deployed an independent auditing system to monitor political ads on Meta in Brazil. To do that we first adapted a browser plugin to gather ads from the timeline of volunteers using Facebook. We managed to convince more than 2,000 volunteers to help our project and install our tool. Then, we use a Convolution Neural Network (CNN) to detect political Meta ads using word embeddings. To evaluate our approach, we manually label a data collection of 10k ads as political or non-political and then we provide an in-depth evaluation of proposed approach for identifying political ads by comparing it with classic supervised machine learning methods. Finally, we deployed a real system that shows the ads identified as related to politics during the 2018 National Brazilian elections. We also investigated early electoral advertisement before the 2020 local Brazilian elections using our model on unsponsored content (regular posts in groups and pages). We noticed that not all political ads we detected were present in the Meta Ad Library for political ads on 2018. Additionally, we found possible early electoral advertisements in 2020, which is forbidden in Brazil. Our results emphasize the importance of enforcement mechanisms for declaring political ads and the need for independent auditing platforms.

The fragmentation of public data in Brazil, coupled with inconsistent standards and limited interoperability, hinders effective research, evidence-based policymaking and access to data-driven insights. To address these issues, we introduce Brazil Data Commons, a platform that unifies various Brazilian datasets under a common semantic framework, enabling the seamless discovery, integration and visualization of information from different domains. By adopting globally recognized ontologies and interoperable data standards, Brazil Data Commons aligns with the principles of the broader Data Commons ecosystem and places Brazilian data in a global context. Through user-friendly interfaces, straightforward query mechanisms and flexible data access options, the platform democratizes data use and enables researchers, policy makers, and the public to gain meaningful insights and make informed decisions. This paper illustrates how Brazil Data Commons transforms scattered datasets into an integrated and easily navigable resource that allows a deeper understanding of Brazil's complex social, economic and environmental landscape.

WhatsApp has become one of the main channels for disseminating health information in Brazil, especially during the COVID-19 pandemic. In this work, we investigate the spread of (mis)information about vaccines in public groups on the platform, focusing on identifying dominant narratives, engagement patterns, and connections to other online information sources. Using a set of 84,640 messages collected between 2020 and 2023, we applied Natural Language Processing (NLP) techniques, including topic modeling (BERTopic) and textual similarity analysis. Overall, the results show that alarmist, misinformative, and politicized content is prevalent, with many messages repeated across multiple groups. We also observed a concentration of activity among a few highly active users, suggesting coordinated amplification strategies. This analysis helps to understand the mechanisms of misinformation spread in private messaging environments and provides insights for public policy and digital health countermeasures.

Political sponsored content has become a powerful yet potentially harmful campaigning tool on Online Social Networks (OSNs). Concerned about its misuse during the 2018 Brazilian elections, we developed a computational framework to audit targeted political advertising. Using a browser extension, we collected ads from over 2,000 Facebook volunteers and trained a Convolutional Neural Network with word embeddings, evaluated against classical machine learning methods on a labeled dataset of 10,000 ads. Our findings reveal gaps in Facebook’s Ad Library and show that coordinated posts on Facebook and Twitter can amplify political messaging, underscoring the need for independent auditing systems.

Although Natural Language Processing (NLP) has traditionally relied on inherently interpretable “white-box” techniques, such as rule-based algorithms, decision trees, hidden Markov models, and logistic regression, the adoption of Large Language Models (LLMs) and language embeddings (often considered “black-box”) has significantly reduced interpretability. This lack of transparency introduces considerable risks, including biases, which have become a major concern in the field of Artificial Intelligence (AI). This Ph.D. thesis addresses these critical gaps by proposing new resources that ensure interpretability and fairness in NLP models for automated fact-checking and hate speech detection tasks. Specifically, we introduce five benchmark datasets (HateBR, HateBRXplain, HausaHate, MOL, and FactNews), four novel post-hoc and self-explaining methods (SELFAR, SSA, B+M and SRA), and one web platform (NoHateBrazil) designed to improve the interpretability and fairness of hate speech detection. The proposed models outperform the existing baselines for Portuguese and Hausa, both underrepresented languages. This research contributes to ongoing discussions on responsible and explainable AI, bridging the gap between model performance and interpretability to achieve positive real-world social impact. Finally, this thesis has had a significant impact both nationally and internationally, receiving citations from prestigious universities and research institutes abroad, inspiring new M.Sc. and Ph.D. in Brazil, and being recognized with multiple awards, including the Google LARA, the Maria Carolina Monard AI Award, and as a finalist at the Brazilian Computer Society Thesis and Dissertation Award.

Misinformation and hate speech form a socially harmful cycle. Research shows that misinformation can amplify hate speech targeting social identity groups and reinforce harmful stereotypes. To combat this cycle, a wide range of Natural Language Processing (NLP) methods have been proposed. Nevertheless, while NLP has historically relied on inherently explainable “white-box” techniques, such as rule-based algorithms, decision trees, hidden markov models, and logistic regression, the adoption of Large Language Models (LLMs) and language embeddings (often considered “black-box”) has significantly reduced interpretability. This lack of transparency introduces considerable risks, including biases, which have become a major concern in AI. This Ph.D. thesis addresses these critical gaps by proposing new resources that ensure explainability and bias mitigation in NLP models for these tasks. Specifically, it introduces five benchmark datasets (HateBR, HateBRXplain, HausaHate, MOL, and FactNews), three novel methods (SELFAR, SSA, and B+M), and one web system (NoHateBrazil) designed to improve the explainability and fairness of automated fact-checking and hate speech detection. The proposed models outperform existing baselines for Portuguese and Hausa, both underrepresented languages. This research contributes to ongoing discussions on responsible and explainable AI, bridging the gap between model performance and interpretability for realworld applications. Finally, it has had a significant impact both nationally and internationally, receiving citations from prestigious universities and research institutes abroad, and inspiring new M.Sc. and Ph.D. projects in Brazil.

The growth of social networks has raised concerns about the misuse of these platforms by disinformation campaigns, social bots, and coordinated activities. Among these platforms, WhatsApp has become a focal point for this abuse, particularly in Brazil, one of the countries with the highest use of the platform. Despite acknowledging the presence of coordinated campaigns and implementing restrictions on the number of messages forwarded per user, the platform continues to be abused. Due to its private nature and the difficulty of collecting information, little is known about these campaigns and the messages they disseminate. Given this context, our study investigates the presence of coordinated activities on WhatsApp in Brazil, identifying their content and purpose, especially how these messages relate to recent Brazilian political events. To answer these questions, we analyzed 13 million messages from 1,444 political groups over seven months from July 2022 to January 2023. Using network analysis, our findings suggest a significant prevalence of coordinated activity in the propagation of news messages, 26% of which originate from misinformation sites. Furthermore, we found that images play a key role in coordinated activity, accounting for 15% of messages, which are also used to mislead. Finally, coordinated accounts were used to organize collective actions, including attacks and protests against election results.