PAD, a pre-train, align, and disentangle framework, improves sequential recommenders, especially for cold items, by aligning frozen LLM embeddings with collaborative embeddings and fusing three experts with frequency-aware gating.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Pre-train, Align, and Disentangle: Empowering Sequential Recommendation with Large Language Models
PAD, a pre-train, align, and disentangle framework, improves sequential recommenders, especially for cold items, by aligning frozen LLM embeddings with collaborative embeddings and fusing three experts with frequency-aware gating.