Activism and misinformation on WhatsApp: measurement, analysis, and countermeasures

• Main Author: Philipe de Freitas Melo
• Publication Date: 2022
• Format: Doctoral thesis
• Language: eng
• Source: Repositório Institucional da UFMG
• Download full: http://hdl.handle.net/1843/49506; https://orcid.org/0000-0001-9830-1896
• topic: WhatsApp; Misinformation; Social networks; Information dissemination; Fake News; Politics; Computação – Teses; WhatsApp (Aplicativo de mensagens) – Teses; Desinformação Teses; Redes sociais – Teses; Disseminação de informação – Teses; Fake News – Teses

The Internet has gained an increasing space in our lives. Social media, especially the Instant Messaging Platforms (IMPs), permeate our relationships and the way we communicate. These platforms allow users to communicate both in a more private and personal conversation, as well as engaging in large public chat groups, in which multiple users interact with each other and share interests. Due to their great success and popularity, these platforms have unfortunately also become a fertile environment for misinformation campaigns that spread rumors, false news, hate speech, and online lynching. However, the closed operating structure of messaging services, such as WhatsApp and Telegram, makes it difficult to investigate the content that circulates within these platforms. In this thesis, we investigate the WhatsApp ecosystem and the (mis)information that spreads on the platform. This task presents a unique challenge, as the content circulating on the WhatsApp network is protected by an end-to-end encryption layer and takes place in private or hard-to-access channels, while, at the same time, this generally anonymous content still manages to spread virally over the network to thousands of users. We focus on this public side of WhatsApp, providing an in-depth understanding of the instant messaging ecosystem, revealing how misinformation campaigns abuse this cyberspace, and also proposing countermeasures to address this issue. First, we developed a new WhatsApp data collection methodology capable of gathering large-scale and long-term data from public groups, in which we collect more than 10 million messages from political groups on WhatsApp in Brazil for more than 3 years. This methodology opens up a new path for studies investigating the WhatsApp platform, revealing trends and the most popular content posted on it. Then, we performed a comparative analysis of IMPs, providing a better understanding of this universe of public groups shared on social networks. We show that there is a substantial amount of public WhatsApp groups and shed light on the main characteristics of the most used instant messaging services. We also explore the behavior of WhatsApp's public political groups, investigating their published messages and some events captured by our dataset, including the misinformation shared during these periods. In relation to these groups, we also explore how misinformation spreads through the WhatsApp network, evaluating how some features of the system impact the flow of information dissemination. Our results reveal that these features help to slow down the spread of content, but do not prevent it from going viral. Finally, we present two practical and viable measures to combat misinformation on WhatsApp. We propose an architecture compatible with WhatsApp E2E encryption and that does not violate any users' privacy. This approach relies on using content that has been previously checked out by specialized agencies and could be easily deployed into the WhatsApp app to help users identify misinformation in the chat. Second, we developed and launched a real system - WhatsApp Monitor - that processes the most popular content shared every day in WhatsApp public groups, ranking and displaying the most uploaded images, videos, audios and images in a web interface that it is already used by many journalists and researchers to fight misinformation.