MindRec: Mind-inspired Coarse-to-fine Decoding for Generative Recommendation

2025-11-16

AI Summary

Existing generative recommender systems rely on autoregressive modeling and greedy decoding, which often trap them in local optima and lack global optimization capability. To address this, we propose MindRec—the first framework to integrate human top-down cognitive reasoning into generative recommendation. MindRec employs keyword-triggered intuitive coarse filtering, leverages a hierarchical category tree to guide fine-grained generation, and introduces Diffusion Beam Search—a novel decoding algorithm that mitigates the locality bias inherent in standard beam search. Built upon large language models (LLMs), MindRec enables human-like, coarse-to-fine recommendation generation. Extensive experiments across multiple benchmarks demonstrate that MindRec achieves an average 9.5% improvement in Top-1 accuracy over state-of-the-art methods. The implementation is publicly available.

Technology Category

• Natural Language Processing: Generation
• Data Mining & Knowledge Management: Recommender Systems
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions

Abstract

Recent advancements in large language model-based recommendation systems often represent items as text or semantic IDs and generate recommendations in an auto-regressive manner. However, due to the left-to-right greedy decoding strategy and the unidirectional logical flow, such methods often fail to produce globally optimal recommendations. In contrast, human reasoning does not follow a rigid left-to-right sequence. Instead, it often begins with keywords or intuitive insights, which are then refined and expanded. Inspired by this fact, we propose Mind-inspired Recommender (MindRec), a novel generative framework that emulates human thought processes. Particularly, our method first generates key tokens that reflect user preferences, and then expands them into the complete item, enabling flexible and human-like generation. To further emulate the structured nature of human decision-making, we organize items into a hierarchical category tree. This structure guides the model to first produce the coarse-grained category and then progressively refine its selection through finer-grained subcategories before generating the specific item. To mitigate the local optimum problem inherent in greedy decoding, we design a novel beam search algorithm, Diffusion Beam Search, tailored for our mind-inspired generation paradigm. Experimental results demonstrate that MindRec yields a 9.5% average improvement in top-1 recommendation performance over state-of-the-art methods, highlighting its potential to enhance recommendation accuracy. The implementation is available via https://github.com/Mr-Peach0301/MindRec.

Problem

• Overcoming left-to-right greedy decoding limitations in LLM-based recommendation systems
• Generating globally optimal recommendations through human-inspired coarse-to-fine reasoning
• Addressing local optimum problems in auto-regressive recommendation generation

Innovation

• Generates key tokens reflecting user preferences first
• Uses hierarchical category tree for coarse-to-fine selection
• Implements Diffusion Beam Search to avoid greedy decoding