Uncategorized

TWON report: Defining metrics for democratic online discourse

Our researcher Sjoerd Stolwijk recently published a deliverable, proposing a set of metrics to determine the deliberative quality of discussions on social media in general, and TWON in particular.

The report lists the key indicators of: (1) Exposure to political content, (2) Engagement with political content, (3) Contributing political content, (4) Diversity of exposure and (5) Quality of exposure.

It is then explained how and why this set of indicators differs from the typical list of deliberative indicators and proposes to view deliberation from a summative rather than an additive perspective. In this view, social media do not need to aim at perfect deliberation within one platform; rather, the goal is to contribute to deliberation at a societal scale via the platform. 

We propose that social media can contribute especially by offering an avenue for users (citizens, journalists and politicians alike) to be exposed to political debate, but also to engage and participate in that debate. In addition, social media can connect otherwise unconnected users and expose them to ideas they might otherwise have missed. Ideally, these ideas are substantiated with arguments and evidence.

Our researcher evaluates a large set of automatic classifiers to determine the degree to which social media comments meet several deliberative criteria, specifically whether comments are rational, interactive, diverse and civil.

Results show how more modern techniques like fine-tuned transformers and generative large language models have improved our ability to reproduce manual codings automatically, but also that results vary considerably between models.

We then integrate the aims of Chapter 3 with the results of Chapter 4 and translate them to the case of TWON to arrive at the metrics proposed in Chapter 2. It adds tests of the performance of different classifiers to determine whether a comment is political or not.

Finally, we take a look into the future, beyond what is currently feasible for TWON, to explore whether new techniques can help determine the deliberative quality of online social media debates to the more fine-grained level of specific claims and show some promising first results.

Download the Deliverable here.

On Regulating Online Social Networks: TWON Policy Brief #2

In January 2025, the TWON consortium developed a second TWON policy brief, on regulating online social networks! The briefing was developed in a comprehensive process with academic input, was then enriched with citizens’ perspectives from the DialoguePerspectives Citizen Lab in Fall 2024 and reviewed multiple times by academics in the consortium. 

The briefing focuses on funding research and the development of public platforms, promoting content diversity through algorithmic design, platform regulation to strengthen interoperability and transparency of platforms, as well as the promotion of media literacy and support of independent journalism.

See the full briefing below &

Download the TWON Policy Brief #2 here

The DSA’s research access: a flawed system

The Digital Service Act does not do enough for research access. While its article §40 implements a duty for very large online platforms (VLOPs) and very large online search engines (VLOSEs) to allow research access, it is not sufficient.
That is why the TWON collaborated with Digits EU, and the Digital Law Institute Trier to feedback the commission on loopholes in this paragraph. The result is an official statement, that the commission now considers.

Their main points of criticism are:

Data access to what? Allow specific requests, as data on minor algorithm changes for specific user groups is currently enclosed. Broaden the definition of systemic risks and create transparency around A/B tests, as they hold great research value.

Data access for whom? Peer review is a standard research process, so peer access to data sets is necessary. Also, allow short-notice data-reaccess to react after peer review, enable group-verifications and create a clear definition of the researcher status to prevent too high barriers.

Verification of data? Currently there is no control mechanism to ensure that the provided data by VLOPs and VLOSEs is correct. An obligation to provide correct datasets needs to be implemented.

Find the full statement here.

Call for Papers on Semantic Generative AI

Call for Papers! 📝🚨

Are you researching social interactions of any combination of agents and/or humans on the web, including the architectures and platforms enabling and influencing those interactions? Then you are a fit.💡

For our workshop on semantic generative agents on the web at the ESWC 2025 in Slovenia we are still accepting papers. Especially welcome are submissions on agent knowledge representation, their reasoning capabilities and way of communication. 💬

Topics of interest include
– Agents for simulating (nonrational) human behavior
– Agents on the (social) web for analyzing communicative behavior
– Platforms for simulating and researching agent communication and platform mechanics
– Recursive AI agents for higher levels of task complexity, adaptivity, and autonomy

We encourage papers from different backgrounds, various disciplines and of various lengths. Papers going beyond traditional paper formats, like demo, data and position papers, are welcome! 🎯

📍Submission deadline is the 6th of March.
📍Paper submission requires the conference attendance of at least one involved researcher. The conference takes place from June 1st to June 5th in Portoroz, Slovenia.

Find out more about it here.

Recommender systems -Shaping our news?

A speech by Damian Trilling​

How can we ensure trust and responsibility in the use of generative AI? What role does it hold in the future of media? What are recommender systems and how do they shape our news feeds?

These were the key questions discussed at the Media Futures Annual Meeting in Bergen, Norway, which brought together over 130 participants from organizations such as @BBC Verify, TV 2, Nokia Bell Labs Cambridge, Amazon Web Services, Vrije Universiteit Amsterdam and Dataminr.
TWON-consortium leader Damian Trilling, professor at Vrjje Universiteit Amsterdam, discussed one of the above mentioned questions in his closing speech: He outlined the impact of different responsible recommender systems for news, as there needs to be a balance between user engagement and ethical considerations.
The Twin of Online Social Networks (TWON) can help to find out, what this balance shall look like and what the impacts of a disbalance are.

The event also featured insightful keynotes that discussed emerging ethical challenges and technological strategies to tackle them, assessed the potential of AI in disaster response, stressed the addressing of falsehoods in uncertain times or showed the impact of regular and early smartphone usage on our world view.

We are happy about the success of the SFI MediaFutures Annual Meeting 2024, fostering collaboration between researchers, industry experts and practitioners.

How to counter Antisemitism online – The European Practitioners network against Antisemitism meets

How does antisemitism appear on social media? What is the role of platform operators when it comes to the moderation of hateful content? And how to counter antisemitism online?

These were only some of the questions discussed during DialoguePerspective’s European Practitioners Network Against Antisemitism meeting in Belgrade this November. In a number of input sessions and workshops participants exchanged the challenges and developed possible solutions, with a focus on intersectionality and younger generations. A visit to a local synagogue gave the chance to grasp realities of Jewish life in Serbia.

In the panel discussion TWON’s Jonas Fegert discussed with Katharina von Schnurbein (European Commission’s Coordination on combating antisemitism and fostering Jewish life), Friedrich Enders (TikTok) and Misko Stanisicšić (Terraforming). Moderated by Jo Frank (DialoguePerspectives) they talked about countering antisemitism online. Jonas Fegert stressed the importance of understanding platform mechanisms, as there is currently a lack of transparency. A Twin of Online Social Networks (a TWON), such as the one we are currently building, is a crucial tool to understanding the effects of platform design choices. The researcher access to platform data guaranteed by the Digital Services Act (DSA) is a chance to shed light in this field. But it is also the responsibility of platform operators to create transparency and cooperate with researchers.

The event was kicked off by remarks by Anke Konrad (German Ambassador to Serbia), Avivit Bar-Ilan (Israeli Ambassador to Serbia), Edward Ferguson (UK Ambassador to Serbia), Brankica Janković (Commission for Protection of Equality, Serbia) and Katharina von Schnurbein.

Thank you for the great discussions with practitioners, activists, policy makers and platform operators!

The first TWON Citizen Lab took place in Karlsruhe!

As part of the Days for Democracy Karlsruhe the participants of DialoguePerspectives’ European Leadership Workshop discussed on how online social networks influence democracy in the digital age.

Achim Rettinger (Trier University), Neta-Paulina Wagner (DialoguePerspectives) & Damian Trilling (Amsterdam University) (from left to right)

In workshops by TWON’s Prof. Damian Trilling, Prof. Achim Rettinger and Dr. Eugen Pissarskoi the participants learned about research approaches to social media dynamics, the ethics behind creating Twins of Online Social Networks (TWON) and the role of online social networks and AI in digital democracy. Melanie Döring and Marisa Wengeler from Institute for Strategic Dialogue (ISD) held thought-provoking inputs on the EU’s Digital Services Act and media literacy as well as the concept of pre- and de-bunking in the online realm. Nuriyatul Lailiyah gave interesting insights on the usage of social media in Southeast Asia. The participants of the European Leadership Workshop not only discussed on the topics within the group, but also with citizens during public outdoor events in the Triangel Event Space Karlsruhe during a Bar Camp and a World Café format.

Discussions in small groups among participants of the European Leadership Workshop

At the end of the 4-day workshop (16.-19. September 2024) the participants worked on a policy brief on necessary regulatory action. This will be used as a basis for further discussion for the scientists of the TWON consortium.

Presentation of ideas for regulatory measures

The European Leadership Workshop moderated by DialoguePerspectives’ programme manager Neta-Paulina Wagner is an integral part of the TWON project:

– It works on the translation from research in to actionable solutions.

– It equips participants with the skills and insights needed to navigate and address the challenges posed by digital democracy.

– Participants actively contribute to the broader conversation on digital citizenship and how we can shape a more resilient, informed, and participatory society in the face of rapidly evolving digital challenges.

A big thank you to our partner DialoguePerspectives for organizing the workshop and thank you to all participants for the great discussions!

Panel Discussion Recap: The Intersection of Generative AI and Online Social Networks

In a panel discussion hosted by AlgorithmWatch in Berlin, the FZI’s Jonas Fegert emphasized the inseparable link between Generative Artificial Intelligence (AI) and Online Social Networks (OSNs). He argued that to fully understand platform mechanisms, we need to develop Digital Twins of Online Social Networks (TWONs), which will provide greater insight into how these platforms function and influence user interactions. With our joint research project, we are taking important stapes in this direction.

The discussion explored the growing role of Generative AI in shaping online communication and the importance of transparent oversight to address the ethical and societal implications of these technologies. In the light of multiple state elections in Germany, the immediate political effects of OSNs and AI were another focal point of the discussion.

Project Review Meeting with the EU-Commission in Brussels

On June 4th the first project review meeting of our EU-funded project “TWON – Twin of Online Social Networks” took place in Brussels. This was a great opportunity to reflect the first year of work in the consortium. We received valuable and encouraging feedback by Prof. Kalina Bontcheva (University of Sheffield), José Manuel Gómez Pérez (Director Language Technology Research at expert.ai) and Ilona von Bethlenfalvy (European Commission), which we will use in the next steps of our project.

On the previous day we also had a consortium meeting in the representation of the state of Baden-Württemberg to the European Union in Brussels – thank you for hosting us!

Jonas Fegert, Michael Mäs, Marko Grobelnik, Achim Rettinger and Damian Trilling (left to right)

Marko Grobelnik, Achim Rettinger, Michael Mäs, Damian Trilling, Alisa Kerschbaum
and Jonas Fegert (left to right)

Achim Rettinger joins working group on AI and its implications on Information and Communication spaces

Since 2024, Achim Rettinger is a member of the Policy Working Group on Artificial Intelligence and its Implications for the Information and Communication Space. The international team focuses on the impact of rapid AI advancements on the global information ecosystem.

For this the team is developing a policy framework aiming to ensure the responsible and democratic use of AI in the information sector. The recommendations address four key areas.

  1. Development and Deployment of AI systems
  2. Accountability regimes (for developers, deployers, users and subjects of AI systems)
  3. Ethical incentives (during development, deployment and use)
  4. Governance of AI

Through this work, the group aims to ensure that AI remains a public good, governed by democratic principles that protect the integrity of information systems and democratic values.