Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MOReGIn: Multi-Objective Recommendation at the Global and Individual Levels

Elizabeth Gómez, David Contreras, Ludovico Boratto, Maria Salamó

TL;DR

The paper tackles the lack of joint global and individual objectives in multi-objective recommender systems by introducing MOReGIn, a post-processing re-ranking method that enforces provider fairness across continents while calibrating recommendations to user genre propensities. By constructing continent-genre buckets and applying a three-phase re-ranking, MOReGIn achieves near-zero disparate visibility and minimized miscalibration, while largely retaining recommendation quality as measured by NDCG. The approach is validated on extended MovieLens-1M data and a novel BeyondSongs dataset, across five baseline algorithms, demonstrating tangible benefits for both providers and users. The work highlights the practicality of integrating global fairness and local calibration in a single framework and points to future directions in exploring consumer fairness and alternative objective combinations.

Abstract

Multi-Objective Recommender Systems (MORSs) emerged as a paradigm to guarantee multiple (often conflicting) goals. Besides accuracy, a MORS can operate at the global level, where additional beyond-accuracy goals are met for the system as a whole, or at the individual level, meaning that the recommendations are tailored to the needs of each user. The state-of-the-art MORSs either operate at the global or individual level, without assuming the co-existence of the two perspectives. In this study, we show that when global and individual objectives co-exist, MORSs are not able to meet both types of goals. To overcome this issue, we present an approach that regulates the recommendation lists so as to guarantee both global and individual perspectives, while preserving its effectiveness. Specifically, as individual perspective, we tackle genre calibration and, as global perspective, provider fairness. We validate our approach on two real-world datasets, publicly released with this paper.

MOReGIn: Multi-Objective Recommendation at the Global and Individual Levels

TL;DR

The paper tackles the lack of joint global and individual objectives in multi-objective recommender systems by introducing MOReGIn, a post-processing re-ranking method that enforces provider fairness across continents while calibrating recommendations to user genre propensities. By constructing continent-genre buckets and applying a three-phase re-ranking, MOReGIn achieves near-zero disparate visibility and minimized miscalibration, while largely retaining recommendation quality as measured by NDCG. The approach is validated on extended MovieLens-1M data and a novel BeyondSongs dataset, across five baseline algorithms, demonstrating tangible benefits for both providers and users. The work highlights the practicality of integrating global fairness and local calibration in a single framework and points to future directions in exploring consumer fairness and alternative objective combinations.

Abstract

Multi-Objective Recommender Systems (MORSs) emerged as a paradigm to guarantee multiple (often conflicting) goals. Besides accuracy, a MORS can operate at the global level, where additional beyond-accuracy goals are met for the system as a whole, or at the individual level, meaning that the recommendations are tailored to the needs of each user. The state-of-the-art MORSs either operate at the global or individual level, without assuming the co-existence of the two perspectives. In this study, we show that when global and individual objectives co-exist, MORSs are not able to meet both types of goals. To overcome this issue, we present an approach that regulates the recommendation lists so as to guarantee both global and individual perspectives, while preserving its effectiveness. Specifically, as individual perspective, we tackle genre calibration and, as global perspective, provider fairness. We validate our approach on two real-world datasets, publicly released with this paper.
Paper Structure (15 sections, 4 equations, 2 figures, 3 tables, 2 algorithms)

This paper contains 15 sections, 4 equations, 2 figures, 3 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Preliminaries
  4. Recommendation Scenario
  5. Metrics
  6. Matching Item Providers and Genre Propensity
  7. Real-world Datasets
  8. Characterizing Group Representation and Genre Propensity
  9. Individually Calibrated and P-Fair Recommendation
  10. Algorithm
  11. Experimental Evaluation
  12. Experimental Methodology
  13. Assessment of Disparities and Mitigation
  14. Impact on the Quality of Recommendations
  15. Conclusions and Future Work

Figures (2)

  • Figure 1: Group representation (a and b) and genre propensity (c and d) in the Movies and Songs data. Acronyms stand for AF: Africa, AS: Asia, EU: Europe, NA: North America, OC: Oceania, SA: South America.
  • Figure 2: Disparity mitigation per continent (a) and miscalibration per genre (b) in BPRMF.

Theorems & Definitions (2)

  • definition 1: Disparate visibility
  • definition 2: Miscalibration