ABSTRACT / People increasingly forge social ties in digital contexts, access and share information, learn, and make decisions such as e.g., what content to watch, what items to buy, or to which job to apply. This has generated an unprecedented opportunity to collect and analyze digital behavioral data, study sociocultural phenomena, and support human activities with computational methods at scale. Information access systems such as recommender systems have become incredibly useful tools to guide users through vast amounts of digital content. However, their trustworthiness is under continuous debate as the underlying algorithms neglect cognitive and social factors of human behavior, can create, or amplify harmful biases, and drive polarization. In this talk, the speaker will present ongoing research on modeling human behavior leveraging psychological models and data-driven recommender systems. The talk will furthermore discuss questions related to fairness and privacy of recommender systems and highlight recent work in this respect.
BIO / Elisabeth Lex is an associate professor and principal investigator of the Recommender Systems and Social Computing Lab at Graz University of Technology (TUG). Her research interests include recommender systems, user modeling, information retrieval, natural language processing and computational social science. She is currently a track chair for User Modeling and Personalization at The 2023 ACM Web Conference and has recently given a tutorial on Trustworthy Algorithmic Ranking Systems at the 16th ACM International WSDM Conference 2023.