Reproduce “SAFERec: Self-Attention and Frequency Enriched Model for Next Basket Recommendation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Oleg Lashinin, Denis Krasilnikov, Aleksandr Milogradskii, Marina Ananyeva
Researchers from Moscow Institute of Physics and Technology, HSE, and T-Bank developed SAFERec, a self-attention and frequency-enriched model for Next Basket Recommendation. This model integrates item frequency information into a Transformer architecture, leading to improved recommendation accuracy, including up to an 8% increase in Recall@10, and an enhanced ability to recommend novel items for entire purchase baskets.
Key insights
- Integrating item frequency information into Transformer architectures is crucial for enhancing their performance in Next Basket Recommendation (NBR) tasks.
- Standard Transformer models, like SASRec, struggle to effectively promote previously consumed and often repeated items in NBR without explicit frequency awareness.
- Combining sophisticated sequential modeling with frequency-aware components allows for a balance between recommendation accuracy and the introduction of novel items.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.