Основні поняття
Explicitly modeling user intentions in within-basket recommendation improves performance significantly.
Анотація
The paper introduces the Neural Pattern Associator (NPA) model for within-basket recommendation, addressing complex user behaviors. NPA encodes user intentions as quantized representations for coherent recommendations. Evaluations show NPA outperforms existing solutions by 5%-25%. The VQA module in NPA identifies combination patterns and contexts efficiently. Strategies like Greedy Search, Weighted Average, and Sampling enhance recommendation accuracy.
Статистика
"NPA model significantly outperforms a wide range of existing WBR solutions."
"Quantitative evaluations show that the NPA model consistently outperforms the baseline models on all three datasets with 5%-25% performance improvement."
"The proposed model exhibits remarkable self-interpretability by tagging in-basket items as recommendation explanations without any post-processing interpretation efforts."
Цитати
"Recent work has shifted its focus toward more practical settings, specifically by incorporating user intentions into the modeling process."
"The NPA model significantly outperforms a wide range of existing WBR solutions, reflecting the benefit of explicitly modeling complex user intentions."