Ensembles of choice-based models for recommender systems

Abstract

In this thesis, we focused on three main paradigms: Recommender Systems, Decision Making, and Ensembles. The work is structured as follows. First, the thesis analyzes the potential of choice-based models. The motivation behind this was based on the idea of applying sound decisionmaking paradigms, such as choice and utility theory, in the field of Recommender Systems. Second, this research analyzes the cognitive process underlying choice behavior. On the one hand, neural and gaze activity were recorded experimentally from different subjects performing a choice task in a Web Interface. On the other hand, cognitive were fitted using rational, emotional, and attentional features. Finally, the work explores the hybridization of choice-based models with ensembles. The goal is to take the best of the two worlds: transparency and performance. Two main methods were analyzed to build optimal choice-based ensembles: uninformed and informed. First one, two strategies were evaluated: 1-Learner and N-Learners ensembles. Second one, we relied on three types of prior information: (1) High diversity, (2) Low error prediction (MSE), (3) and Low crowd error.