Show & Tell Report | 18. March 2025

Workshop Report: Lecture and Discussion Series "Social media data: Show & Tell - Right-wing extremism and democracy studies (1)" (Online)

By Dr. Vincent Fröhlich

Since early 2022, the working group 'Social Media Data,' an initiative of NFDI4Culture in collaboration with BERD@NFDI, KonsortSWD and Text+ within the framework of the National Research Data Infrastructure, has been organizing the lecture and discussion series "Show & Tell – Social Media Data in Research Practice."

Each session, lasting 90 minutes on Zoom, focuses on tools used in social media research and highlights best practices from selected projects. The series explores not only pragmatic solutions and technical possibilities (interfaces, repositories, metadata standards, interoperability, etc.) but also ethical and legal challenges (e.g., personal rights, copyright) in the sustainable, secure, and critical handling of such data (code and data literacy, FAIR & CARE principles). Additionally, the series fosters discussions on interdisciplinary research approaches and teaching methods that challenge traditional and field-specific frameworks and tools.

In 2025, the series will focus specifically on handling social media data in research on right-wing extremism and democracy.

First Session on February 28, 2025:

The inaugural session featured Christian Donner from Institut für Demokratie und Zivilgesellschaft (IDZ) und Pascal Siegers von DP-R|EX, the Data Portal for Research on Racism and Right-Wing Extremism.

Christian Donner introduced the core product of his project and its predecessor: Machine Against the Rage, an online magazine for digital conflict research that provides data analyses on hate speech and disinformation. His project monitors anti-democratic actors by analyzing social dynamics and their escalation, defining this work as digital conflict research. The primary platform under study is Telegram, but data from YouTube, TikTok, and previously Twitter are also analyzed. Donner provided an in-depth overview of the Telegram data analysis.

The project monitors 2,000 Telegram channels, from which data have been collected weekly since September 2022. This includes text, images, videos, links, subscriber counts, views (biweekly), and reactions. The channels are categorized into three main groups: right-wing extremism, conspiracism, and others (preppers, Russian imperialism, and pro-Russian propaganda).

Data were collected using snowball sampling and are frequently compared with earlier samples. Selection criteria for monitored channels include having at least 500 subscribers, being German-language, and being public. The project also offers external access and partnerships with other research projects, such as studies on the Querdenker movement and civil society organizations lacking in-house data expertise. A key principle in these collaborations is data minimization.

Findings derived from the data include information on locations and social networks, the relevance and growth of groups, and the timeliness and popularity of posts. These insights are published in RADAR, an analysis of metric trends within the monitored sample over the last three months, highlighting specific themes such as mobilization, monetization, images/memes, and AI-generated images.

Additionally, Donner briefly introduced YouTube data analysis. Data is accessed via the official API and external APIs, capturing video titles, descriptions, channel information, view and like statistics, and transcripts (for one-third of the videos). The project also examines cross-linkages between Telegram and YouTube. TikTok data analysis is still under development.

Pascal Siegers described how the Federal Ministry of Education and Research (BMBF) requested several years ago that efforts be made to improve infrastructure for right-wing extremism research (abbreviated as "Rex research"). The goal was to reduce redundant data collection by enhancing data visibility and promoting data sharing. A pilot study revealed that 80% of the data collected in this field over the previous five years had not been shared. The project aims to facilitate research in this area.

Developed solutions include tools for data search, data sharing, research data management (RDM) support, a hate speech corpus, and a data trust system (AVERA), which is still under development. Siegers focused on the latter two initiatives—the corpus and the data trust system.

The data trust system addresses redundant efforts in creating account lists for actor-based social media data collection. Although legally feasible, data sharing is complex. The initiative seeks to establish a community-based data trust to solve three problems:

1. A trust issue (minimizing legal risks),

2. A collective goods issue (creating incentives), and

3. A data quality issue (social media as a challenge).

The data trust system would facilitate a shared data pool, with the trust administrator ensuring compliance with data usage regulations. DP-R|EX’s preparatory work is expected to support expansions such as Com-DTM and potentially the Com-DS project.

Kathrin Weller from the GESIS team supplemented Siegers’ presentation by discussing the hate speech research corpus. It remains uncertain whether a text corpus can be made available. However, individual researchers and research groups can benefit from the formal processes being developed, which address key questions such as: How should the architecture for Telegram data research be structured? What security measures need to be planned? How should a data collection pipeline be built?

Beyond these project presentations, the session focused on seven overarching questions regarding the handling of social media data. These discussions and insights will be published on the Knowledge Base.

During the subsequent plenary discussion, concerns were raised regarding researchers in the field of right-wing extremism being particularly exposed. The community noted that universities and research institutions have yet to implement special security measures, leaving researchers without institutional protection.

Upcoming Session:

The next episode of "Social Media Data: Show & Tell – Research on Right-Wing Extremism and Democracy (2)" will take place on Friday, April 25, 2025, at 2:00 PM. Presentations will be given by:

• Prof. Dr. Jasmin Riedl (Universität der Bundeswehr München, Project: SPARTA)
• Ina Ni and Isabel Bezzaoui (FZI: Projects DeFaktS and SOSEC – Social Sentiment in Times of Crises)

Organizers:

• BERD@NFDI – NFDI Consortium for Business, Economic, and Related Data
• KonsortSWD – Consortium for the Social, Behavioral, Educational, and Economic Sciences
• NFDI4Culture – Consortium for Research Data on Tangible and Intangible Cultural Assets
• Text+ – Consortium for Text- and Language-Based Research Data

Coordination:

Vincent Fröhlich (vincent.froehlich@staff.uni-marburg.de)

Philippe Genêt (P.Genet@dnb.de)

If you would like to become part of the Social Media Data Working Group in the NFDI, please subscribe to our mailing list