Don't Waste It: Guiding Generative Recommenders with Structured Human Priors via Multi-head Decoding

2025-11-13

AI Summary

This work addresses the challenge of effectively incorporating structured domain knowledge—i.e., human priors—into end-to-end generative recommender systems. We propose a backbone-agnostic multi-head adapter framework that injects such priors directly into the decoding process via lightweight adapter heads and a hierarchical composition strategy. This enables end-to-end, interpretable disentanglement of user intent along semantic dimensions and modeling of interactions among complex priors—without relying on post-hoc refinement or fine-tuning. Our key innovation lies in grounding human priors explicitly within the generative process itself. Extensive experiments on three large-scale benchmark datasets demonstrate significant improvements across recommendation accuracy, diversity, novelty, and personalization. Moreover, the method enhances utilization of long-context inputs and scales effectively with larger language models.

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: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Optimizing recommender systems for objectives beyond accuracy, such as diversity, novelty, and personalization, is crucial for long-term user satisfaction. To this end, industrial practitioners have accumulated vast amounts of structured domain knowledge, which we term human priors (e.g., item taxonomies, temporal patterns). This knowledge is typically applied through post-hoc adjustments during ranking or post-ranking. However, this approach remains decoupled from the core model learning, which is particularly undesirable as the industry shifts to end-to-end generative recommendation foundation models. On the other hand, many methods targeting these beyond-accuracy objectives often require architecture-specific modifications and discard these valuable human priors by learning user intent in a fully unsupervised manner. Instead of discarding the human priors accumulated over years of practice, we introduce a backbone-agnostic framework that seamlessly integrates these human priors directly into the end-to-end training of generative recommenders. With lightweight, prior-conditioned adapter heads inspired by efficient LLM decoding strategies, our approach guides the model to disentangle user intent along human-understandable axes (e.g., interaction types, long- vs. short-term interests). We also introduce a hierarchical composition strategy for modeling complex interactions across different prior types. Extensive experiments on three large-scale datasets demonstrate that our method significantly enhances both accuracy and beyond-accuracy objectives. We also show that human priors allow the backbone model to more effectively leverage longer context lengths and larger model sizes.

Problem

• Integrating structured human priors into generative recommender training
• Guiding models to disentangle user intent along interpretable axes
• Enhancing both accuracy and beyond-accuracy objectives simultaneously

Innovation

• Integrates human priors into generative recommender training
• Uses lightweight adapter heads to guide user intent
• Employs hierarchical composition for complex prior interactions