SimGR: Escaping the Pitfalls of Generative Decoding in LLM-based Recommendation

2026-02-08

AI Summary

This work addresses a critical yet overlooked issue in existing generative recommendation methods: the systematic bias introduced by erroneously modeling item-level user preferences as token-level autoregressive generation. To resolve this mismatch, we propose SimGR, a novel framework that, for the first time, formally identifies and theoretically analyzes this modeling discrepancy. Instead of relying on conventional beam search or independent token assumptions, SimGR directly models item-level preference distributions within a shared latent space and aligns the generation process with the recommendation objective through similarity-based ranking. Extensive experiments across multiple datasets and diverse large language model backbones demonstrate that SimGR consistently and significantly outperforms current state-of-the-art approaches, confirming its effectiveness and robustness.

Technology Category

• Natural Language Processing: Generation
• Machine Learning: Learning Preferences or Rankings
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Web data generation and simulation

Abstract

A core objective in recommender systems is to accurately model the distribution of user preferences over items to enable personalized recommendations. Recently, driven by the strong generative capabilities of large language models (LLMs), LLM-based generative recommendation has become increasingly popular. However, we observe that existing methods inevitably introduce systematic bias when estimating item-level preference distributions. Specifically, autoregressive generation suffers from incomplete coverage due to beam search pruning, while parallel generation distorts probabilities by assuming token independence. We attribute this issue to a fundamental modeling mismatch: these methods approximate item-level distributions via token-level generation, which inherently induces approximation errors. Through both theoretical analysis and empirical validation, we demonstrate that token-level generation cannot faithfully substitute item-level generation, leading to biased item distributions. To address this, we propose \textbf{Sim}ply \textbf{G}enerative \textbf{R}ecommendation (\textbf{SimGR}), a framework that directly models item-level preference distributions in a shared latent space and ranks items by similarity, thereby aligning the modeling objective with recommendation and mitigating distributional distortion. Extensive experiments across multiple datasets and LLM backbones show that SimGR consistently outperforms existing generative recommenders. Our code is available at https://anonymous.4open.science/r/SimGR-C408/

Problem

• generative recommendation
• large language models
• preference distribution
• token-level generation
• item-level modeling

Innovation

• generative recommendation
• item-level preference
• distributional bias
• latent space modeling
• large language models