Rank-sensitive proportional aggregations in dynamic recommendation scenarios

Abstract

In this paper, we focus on the problem of rank-sensitive proportionality preservation when aggregating outputs of multiple recommender systems in dynamic recommendation scenarios. We believe that individual recommenders may provide complementary views on the user’s preferences or needs, and therefore, their proportional (i.e. unbiased) aggregation may be beneficial for the long-term user satisfaction. We propose an aggregation framework (FuzzDA) based on a modified D’Hondt’s algorithm (DA) for proportional mandates allocation. Specifically, we adjusted DA to register fuzzy membership of items and modified the selection procedure to balance both relevance and proportionality criteria. Furthermore, we propose several iterative votes assignment strategies and negative implicit feedback incorporation strategies to make FuzzDA framework applicable in dynamic recommendation scenarios. Overall, the framework should provide benefits w.r.t. long-term novelty of recommendations, diversity of recommended items as well as overall relevance. We evaluated FuzzDA framework thoroughly both in offline simulations and in online A/B testing. Framework variants outperformed baselines w.r.t. click-through rate (CTR) in most of the evaluated scenarios. Some variants of FuzzDA also provided the best or close-to-best iterative novelty (while maintaining very high CTR). While the impact of the framework variants on user-wise diversity was not so extensive, the trade-off between CTR and diversity seems reasonable.

Preface to the special issue on dynamic recommender systems and user models

“…Rank-sensitive proportional aggregations in dynamic recommendation scenarios (Balcar et al 2022) is an extension of the authors' work presented at the 3rd Workshop on Online Recommender Systems and User Modeling (Peska and Balcar 2019). The contribution consists of FuzzDA, an unbiased online rank aggregation framework that relies on proportionality provided by a modified D'Hont's algorithm for proportional mandates allocation.…”

Hierarchical Portfolios in Recommender Ecosystems

“…As a way of performing aggregation, d'Hondt's algorithm has proved its merits [2]. However, in the cited paper personalisation was performed only during integration of implicit negative feedback (INF).…”

“…Base recommenders recommend 100 items and aggregator returns a list of 20 items. Users' ability to observe item [2] on kth position is simulated in two ways: At first, static P stat08 (noticed|k) = 0.8 or P stat06 (noticed|k) = 0.6 (probability of noticing is equal to 0.8 or 0.6 for all the items) and at second, linear P lin0901 (probability that user has noticed the item decreases from 0.9 to 0.1 linearly with item position in the list).…”