Token-level Collaborative Alignment for LLM-based Generative Recommendation

2026-01-26

AI Summary

This work addresses the challenge of effectively integrating collaborative filtering (CF) signals into large language model (LLM)-based generative recommender systems, which stems from a semantic granularity mismatch between item-level user preferences in CF and token-level prediction objectives in LLMs. To bridge this gap, the authors propose TCA4Rec, a framework that establishes an explicit optimization-level interface between CF and LLM generation. Specifically, a collaborative tokenizer maps item-level CF logits into token-level distributions, which are then aligned with the LLM’s output via a soft label alignment mechanism. This approach enables model-agnostic, plug-and-play token-level collaborative alignment. Extensive experiments demonstrate that TCA4Rec consistently enhances recommendation accuracy across diverse CF models and LLM architectures while preserving textual fluency, thereby validating its generality, effectiveness, and controllability.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Large Language Models (LLMs) have demonstrated strong potential for generative recommendation by leveraging rich semantic knowledge. However, existing LLM-based recommender systems struggle to effectively incorporate collaborative filtering (CF) signals, due to a fundamental mismatch between item-level preference modeling in CF and token-level next-token prediction (NTP) optimization in LLMs. Prior approaches typically treat CF as contextual hints or representation bias, and resort to multi-stage training to reduce behavioral semantic space discrepancies, leaving CF unable to explicitly regulate LLM generation. In this work, we propose Token-level Collaborative Alignment for Recommendation (TCA4Rec), a model-agnostic and plug-and-play framework that establishes an explicit optimization-level interface between CF supervision and LLM generation. TCA4Rec consists of (i) Collaborative Tokenizer, which projects raw item-level CF logits into token-level distributions aligned with the LLM token space, and (ii) Soft Label Alignment, which integrates these CF-informed distributions with one-hot supervision to optimize a soft NTP objective. This design preserves the generative nature of LLM training while enabling collaborative alignment with essential user preference of CF models. We highlight TCA4Rec is compatible with arbitrary traditional CF models and generalizes across a wide range of decoder-based LLM recommender architectures. Moreover, it provides an explicit mechanism to balance behavioral alignment and semantic fluency, yielding generative recommendations that are both accurate and controllable. Extensive experiments demonstrate that TCA4Rec consistently improves recommendation performance across a broad spectrum of CF models and LLM-based recommender systems.

Problem

• Large Language Models
• Generative Recommendation
• Collaborative Filtering
• Token-level Alignment
• Next-token Prediction

Innovation

• Token-level Alignment
• Collaborative Filtering
• Generative Recommendation
• Large Language Models
• Soft Label Alignment