Researchers have developed LSNR, a new approach for learned sparse news recommendations that addresses limitations in current methods. Unlike previous techniques that concatenate user interaction histories into a single query, LSNR encodes each article individually. It then constructs a sparse user vector by using attention-based weighting of these article representations, combined with a dense-projected expansion at the user level. Experiments on the MIND and EBNeRD datasets demonstrate that LSNR surpasses existing methods like STAR, particularly in scenarios with short user histories, and remains competitive with state-of-the-art dense recommenders. AI
IMPACT This new method could improve the efficiency and effectiveness of news recommendation systems by better handling user interaction histories.
RANK_REASON Academic paper detailing a new method for recommender systems. [lever_c_demoted from research: ic=1 ai=1.0]
Read on arXiv cs.IR (Information Retrieval) →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
