Multi-objective pareto-efficient algorithms for recommender systems

Abstract

Recommender systems are quickly becoming ubiquitous in applications such as ecommerce, social media channels and content providers, acting as enabling mechanisms designed to overcome the information overload problem by improving browsing and consumption experience. A typical task in recommender systems is to output a ranked list of items, so that items placed higher in the rank are more likely to be interesting to the users. Interestingness measures include how accurate, novel and diverse the suggested items are, and the objective is usually to produce ranked lists optimizing one of these measures. Suggesting items that are simultaneously accurate, novel and diverse is much more challenging, since this may lead to a conflicting-objective problem, in which the attempt to improve a measure further may result in worsening other measures. In this thesis we propose new approaches for multi-objective recommender systems based on the concept of Pareto-efficiency -- a state achieved when the system is devised in the most efficient manner in the sense that there is no way to improve one of the objectives without making any other objective worse off. Given that existing recommendation algorithms differ in their level of accuracy, diversity and novelty, we exploit the Pareto-efficiency concept in two distinct manners: (i) the aggregation of ranked lists produced by existing algorithms into a single one, which we call Paretoefficient ranking, and (ii) the weighted combination of existing algorithms resulting in a hybrid one, which we call Pareto-efficient hybridization. Our evaluation involves two real application scenarios: music recommendation with implicit feedback (i.e., Last.fm) and movie recommendation with explicit feedback (i.e., MovieLens). We show that the proposed approaches are effective in optimizing each of the metrics without hurting the others, or optimizing all three simultaneously. Further, for the Pareto-efficient hybridization, we allow for adjusting the compromise between the metrics, so that the recommendation emphasis can be set dinamically according to the needs of different users.