Title: Diversity and precision of Bayesian Mallows to learn preferences from clicking data
Authors: Arnoldo Frigessi - University of Oslo (Norway) [presenting]
Abstract: Clicking data contain user preference information and can be used to produce personalized recommendations in web-based applications. We propose the Bayesian Mallows for Clicking Data method, which augments clicking data into compatible full ranking vectors. User preferences are learned using a Mallows ranking model. Bayesian inference leads to interpretable uncertainties of each individual recommendation. With a simulation study and a data example, we demonstrate that compared to state-of-the-art matrix factorization, our method makes personalized recommendations with similar accuracy, while achieving higher level of diversity, and producing interpretable and actionable uncertainty estimation. We discuss computationally efficient model approximations.