Researchers have developed a new framework called soft-token fusion to integrate numerical and embedding features into Large Language Models (LLMs) for recommender systems. This approach addresses the limitation of current LLM-based recommenders that primarily process discrete textual tokens, enabling them to better utilize continuous and dense features. Experiments on Amazon recommendation benchmarks demonstrated that this method improves retrieval performance compared to existing LLM baselines, with an interaction-based fusion module proving more effective than simple concatenation. AI
IMPACT Enhances LLM capabilities in recommender systems by enabling better integration of diverse data types.
RANK_REASON The cluster contains an academic paper detailing a new method for LLMs in recommender systems.
Read on arXiv cs.IR (Information Retrieval) →
- arXiv
- Hugging Face
- interaction-based fusion module
- LLM
- recommender system
- soft-token fusion
- two-tower retrieval model
AI-generated summary · Google Gemini · from 3 sources. How we write summaries →
