Relevance and diversity are critical objectives in modern information retrieval (IR), particularly in recommender systems. Achieving a balance between relevance (exploitation) and diversity (exploration) optimizes user satisfaction and business goals such as catalog coverage and novelty. While existing post-processing reranking methods address this trade-off, they usually rely on greedy strategies, leading to suboptimal outcomes for large-scale tasks. To this end, we propose Sampled Maximal Marginal Relevance (SMMR), a novel sampling-based extension of MMR that introduces randomness into item selection to improve relevance-diversity trade-offs. SMMR avoids the rigidity of greedy and deterministic reranking, and achieves a logarithmic computational speedup, which allows it to scale on large candidate sets. Our evaluations on multiple real-world open-source datasets demonstrate that SMMR consistently outperforms existing state-of-the-art approaches, offering superior performance in balancing relevance and diversity. Our implementation of the proposed method is made available to support future research.

Bayesian Personalized Ranking (BPR), a collaborative filtering approach based on matrix factorization, frequently serves as a benchmark for recommender systems research. However, numerous studies often overlook the nuances of BPR implementation, claiming that it performs worse than newly proposed methods across various tasks. In this paper, we thoroughly examine the features of the BPR model, indicating their impact on its performance, and investigate open-source BPR implementations. Our analysis reveals inconsistencies between these implementations and the original BPR paper, leading to a significant decrease in performance of up to 50% for specific implementations. Furthermore, through extensive experiments on real-world datasets under modern evaluation settings, we demonstrate that with proper tuning of its hyperparameters, the BPR model can achieve performance levels close to state-of-the-art methods on the top-n recommendation tasks and even outperform them on specific datasets. Specifically, on the Million Song Dataset, the BPR model with hyperparameters tuning statistically significantly outperforms Mult-VAE by 10% in NDCG@100 with binary relevance function.

In this paper we study the next basket recommendation problem. Recent methods use different approaches to achieve better performance. However, many of them do not use information about the time of prediction and time intervals between baskets. To fill this gap, we propose a novel method, Time-Aware Item-based Weighting (TAIW), which takes timestamps and intervals into account. We provide experiments on three real-world datasets, and TAIW outperforms well-tuned state-of-the-art baselines for next-basket recommendations. In addition, we show the results of an ablation study and a case study of a few items.