Managing Popularity Bias in Recommender Systems with Personalized Re-ranking
This addresses the problem of under-recommending niche products for businesses in recommender systems, though it appears incremental as a post-processing improvement.
The paper tackles popularity bias in recommender systems by introducing a personalized diversification re-ranking approach that increases representation of less popular items while maintaining acceptable recommendation accuracy, showing it manages bias more effectively than an existing regularization-based method.
Many recommender systems suffer from popularity bias: popular items are recommended frequently while less popular, niche products, are recommended rarely or not at all. However, recommending the ignored products in the `long tail' is critical for businesses as they are less likely to be discovered. In this paper, we introduce a personalized diversification re-ranking approach to increase the representation of less popular items in recommendations while maintaining acceptable recommendation accuracy. Our approach is a post-processing step that can be applied to the output of any recommender system. We show that our approach is capable of managing popularity bias more effectively, compared with an existing method based on regularization. We also examine both new and existing metrics to measure the coverage of long-tail items in the recommendation.