September 26, 2025, SFI MediaFutures proudly participated in the 13th International Workshop on News Recommendation and Analytics (INRA2025), held as part of the prestigious ACM RecSys conference in Prague. The workshop served as a vibrant platform to explore the future of news recommendation systems, with a strong focus on personalization, emotional alignment, humor, and content provenance.
MediaFutures’ team presented four innovative research papers, each exploring new dimensions of news recommendation systems and the role of generative AI in shaping user experiences.
The first presentation, led by research assistant Gloria Kasangu, introduced a longitudinal evaluation study on the effects of personalization in news recommender systems. The study investigates whether users truly benefit from increased personalization, offering early insights into long-term user engagement and satisfaction.
Research assistant Tobias Jovall Wessel followed with a creative exploration of satirical reframing of news articles using large language models (LLMs). His work examines how humor and satire can make news more accessible and engaging, raising important questions about tone and audience perception.
Associate professor Alain D. Starke presented research conducted with Jørgen Eknes-Riple, Jia-Hua Jeng, and Khadiga Seddik on emotional alignment in news recommendations. This study used LLMs to reframe articles to match users’ emotional states, investigating how such alignment influences user choices and engagement.
The final presentation featured PhD candidate Svenja Lys Forstner and research assistant Yelyzaveta Lysova, who showcased a study on image provenance labeling using C2PA technology. Inspired by food packaging labels, their work tested the effectiveness of visual indicators such as “Content Credentials” in helping users identify AI-generated or altered images in news content.
All four projects reflect MediaFutures’ commitment to responsible innovation, user-centric design, and the strategic use of generative AI in media technology research. The work was supervised by Centre Director Christoph Trattner, highlighting the centre’s leadership in advancing trustworthy and impactful media solutions.
MediaFutures continues to push the boundaries of news recommendation research and looks forward to sharing further developments in the near future.
————————————————-
In total six publications were accepted at different workshops co-located with ACM RecSys 2025:
More of the Same? A Longitudinal Evaluation of Two Similarity-based Approaches in a News Recommender System. Kasangu, G., Starke, A., Trattner, C. INRA Workshop. 2025.
Using Large Language Models to ‘Lighten the Mood’: Satirically Reframing News Recommendations to Reduce News Avoidance. Wessel, T., Trattner, C., Starke, A. INRA Workshop. 2025
Evaluating Image Trust Labels in a News Recommender System. Forstner, S. L., Lysova, Y., Starke, A. D., Trattner, C. INRA Workshop. 2025.
Hope, Fear, or Anger? How Emotional Framing in a News Recommender System Guides User Preferences. Eknes-Riple, J., Hua, J., Jeng, J., Starke, A. D., Seddik, K. M. A., Trattner, C. INRA Workshop. 2025.
Nudging Healthy Choices: Leveraging LLM-Generated Hashtags and Explanations in Personalised Food Recommendations. El Majjodi, A., Trattner, C., Starke, A., Petruzzelli, A., Musto, C. IntRS Workshop. 2025.
LLM-based Healthiness? Analyzing the Nutritional Quality of an AI-Generated Recipe Dataset. El Majjodi, A., Trattner, C., Holtan, J. B., Starke, A. RecSoGood Workshop. 2025.