Doctoral colloquium - Elena Štefancová (19.5.2025)
Monday 19.5.2025 at 13:10 hod., Lecture room I/9
Elena Štefancová:
Trustworthy Recommender Systems
Abstract:
Artificial intelligence plays an increasingly important role in our society, however it also entails risks. Some of the potential solutions to make AI more trustworthy are greater transparency and fairness.
This work investigates specific fairness aspects within recommender systems, focusing on their inter-relations and trade-offs. By analyzing various fairness dimensions, the research aims to interpret the results in a manner that enhances users' trust in the outputs provided by these systems.
Furthermore, the work develops a novel synthetic data generation method designed to evaluate changes in the internal representation of recommender systems, facilitating a deeper understanding of their behavior and decision-making processes.
The outcomes of this research provide valuable insights into the balance between fairness and performance, offering tools for more transparent and trustworthy recommender systems.