We have been holding back on new meetups in the hope we could see you all again in real life, but no longer! On Tuesday November 26th, we will host our first ever virtual RecSysNL meetup. We have two exciting speakers and look forward to checking in with you all!
- Talk 1: Toward Natural Language Mitigation Strategies for Cognitive Biases in Recommender Systems - Alisa Rieger, PhD Student, TU Delft
- Talk 2: Blendle: Diverse content recommendation from a vast collection - Jasper Oosterman - Data Scientist
Second abstract and link to be announced soon.
Abstract: Toward Natural Language Mitigation Strategies for Cognitive Biases in Recommender Systems - Alisa Rieger, PhD Student, TU Delft
Decision-making at individual, business, and societal levels is influenced by online news and social media. Filtering and ranking algorithms such as recommender systems are used to support these decisions. Further, individual cognitive selection strategies and homogeneous networks can amplify bias in customized recommendations, and influence which information we are exposed to. A first step in the direction of bias mitigation would consequently be to raise users' awareness of filtering mechanisms and potential cognitive biases, e.g., through explanations and interactive interfaces.
We review studies in the field of recommender systems focused on interface-based mitigation of various cognitive biases. Analysis and comparison of previous studies provides an overview of effective implementation and methodological evaluation approaches for these biases.