Conditional Memory Enhanced Item Representation for Generative Recommendation
cs.IR, cs.AI
Submitted: 2026-05-12
Updated: 2026-10-28
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
The gist: Generative recommendation (GR) has emerged as a promising paradigm that predicts target items by autoregressively generating their semantic identifiers (SID).
Terminology
Abstract
Generative recommendation (GR) has emerged as a promising paradigm that predicts target items by autoregressively generating their semantic identifiers (SID). Most GR methods follow a quantization-representation-generation pipeline. While most research focuses on quantization and generation, the representation that bridges the two remains underexplored. Existing representation constructions mainly take two forms: either directly flattening the SID code embeddings into a sequence or merging them into a compact vector via external networks. While effective, they still expose two practical challenges: 1) Loss of item semantics: Information lost by quantization or compression needs to be recovered during the representation stage. 2) Insufficient modeling of structured SID relations: Intra-item code composition patterns and inter-item prefix-transition patterns are not explicitly preserved for representation and subsequent SID generation. To overcome these challenges, we propose ComeIR, a nditional mory Enhanced tem epresentation framework that reconstructs SID code embeddings into item-aware inputs and preserves structured SID evidence for prediction. Item-guided Token Scoring estimates the contribution of each SID code using the item embedding employed by the quantizer. Dual-level Conditional Memory captures intra-item prefix composition and inter-item prefix transitions, while Memory Merge integrates these signals into each historical item representation. The Memory-restoring Prediction Head further conditions SID-level prediction on the generated prefix and historical transition evidence. Extensive experiments demonstrate the effectiveness and flexibility of ComeIR, and further reveal scalable gains from enlarging conditional memory.
Sources
- Bi-Level Optimization for Generative Recommendation: Bridging Tokenization and Generation
- Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation
- Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models
- OneRec: Unifying Retrieve and Rank with Generative Recommender and Iterative Preference Alignment
- HiD-VAE: Interpretable Generative Recommendation via Hierarchical and Disentangled Semantic IDs
- When Attention Sink Emerges in Language Models: An Empirical View
- Stop Treating Collisions Equally: Qualification-Aware Semantic ID Learning for Recommendation at Industrial Scale
- A Survey of Generative Search and Recommendation in the Era of Large Language Models
- Rec-R1: Bridging Generative Large Language Models and User-Centric Recommendation Systems via Reinforcement Learning
- A Collision-Free Hot-Tier Extension for Engram-Style Conditional Memory: A Controlled Study of Training Dynamics
- Efficient Inference for Large Language Model-based Generative Recommendation
- Generative Recommender with End-to-End Learnable Item Tokenization
- A Parametric Memory Head for Continual Generative Retrieval
- Auxiliary-Loss-Free Load Balancing Strategy for Mixture-of-Experts
- IntRR: A Framework for Integrating SID Redistribution and Length Reduction
- Efficient Streaming Language Models with Attention Sinks
- Unifying Generative and Dense Retrieval for Sequential Recommendation
- Sparse Meets Dense: Unified Generative Recommendations with Cascaded Sparse-Dense Representations
- Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations
- A Survey of Large Language Models
Related papers
- Revisiting Text Ranking in Deep Research
- GateSID: One Gate to Balance Semantic and Collaborative Signals across the Item Lifecycle
- SAM-D2Q: Aligning Multimodal Doc2Query with Search Demand and Conversion for E-commerce
- GINGER: Spicing Up Graph-Based Vector Search
- Subtraction Gets You More: Gap-Aware Retrieval for Multimodal Multi-Hop QA
- One-Step Retrieval Framework for Real-Time Sponsored Search Ads Using Hierarchical Text Representations