CAPRI-FAIR: Integration of Multi-sided Fairness in Contextual POI Recommendation Framework

TL;DR

CAPRI-FAIR tackles multi-stakeholder fairness in context-aware POI recommendations by adding two post-filter fairness factors to the CAPRI framework. It systematically evaluates linear, power-law, and logistic provider fairness models and a consumer fairness strategy across Yelp and Gowalla datasets, reporting results with Precision@k and Generalized Cross-Entropy (GCE) to assess accuracy and exposure fairness. The findings show that the linear provider fairness model improves long-tail exposure with modest precision loss, while consumer fairness can boost inactive-user precision in some cases, with combined fairness producing Pareto tradeoffs that depend on model and dataset. The work demonstrates a practical approach to balancing provider exposure and user relevance in real-world POI systems and highlights dataset/model-dependent tradeoffs that guide deployment choices.

Abstract

Point-of-interest (POI) recommendation considers spatio-temporal factors like distance, peak hours, and user check-ins. Given their influence on both consumer experience and POI business, it's crucial to consider fairness from multiple perspectives. Unfortunately, these systems often provide less accurate recommendations to inactive users and less exposure to unpopular POIs. This paper develops a post-filter method that includes provider and consumer fairness in existing models, aiming to balance fairness metrics like item exposure with performance metrics such as precision and distance. Experiments show that a linear scoring model for provider fairness in re-scoring items offers the best balance between performance and long-tail exposure, sometimes without much precision loss. Addressing consumer fairness by recommending more popular POIs to inactive users increased precision in some models and datasets. However, combinations that reached the Pareto front of consumer and provider fairness resulted in the lowest precision values, highlighting that tradeoffs depend greatly on the model and dataset.

Figures (10)

• Figure 1: Long-tail Exposure v.s. Provider fairness factor $\alpha$
• Figure 2: Precision@10 v.s. Consumer fairness factor $\beta$
• Figure 3: Scatterplot showing the tradeoff between GCE for users, GCE for items, and precision
• Figure 4: Histogram of popularity /check-in counts in the Yelp training dataset, with ridge regression linear model, $\alpha = 10.0$.
• Figure 5: Different provider fairness scoring models.