Existence conditions for hidden feedback loops in online recommender systems
This addresses degradation in coverage and novelty for online recommender systems, but the work is incremental as it builds on known feedback loop issues.
The paper investigates hidden feedback loops in online recommender systems, showing that unbiased additive noise does not prevent loops, while a non-zero reset probability can limit them, with experiments confirming these findings for four bandit algorithms.
We explore a hidden feedback loops effect in online recommender systems. Feedback loops result in degradation of online multi-armed bandit (MAB) recommendations to a small subset and loss of coverage and novelty. We study how uncertainty and noise in user interests influence the existence of feedback loops. First, we show that an unbiased additive random noise in user interests does not prevent a feedback loop. Second, we demonstrate that a non-zero probability of resetting user interests is sufficient to limit the feedback loop and estimate the size of the effect. Our experiments confirm the theoretical findings in a simulated environment for four bandit algorithms.