Time-Aware Diffusion based on Preference Disentanglement for Generative Recommendation

2026-06-01 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Expert, quick

Summary

TDPM, a novel Generative Recommender (GR) framework, addresses a critical limitation in existing diffusion-based GRs by integrating time-evolving user preferences into the diffusion process on semantic index (SID) tokens. Current diffusion GRs apply uniformly to historical interactions, failing to account for the non-stationary nature of user preferences. TDPM disentangles user preference into a long-term consistent "period preference" and a "point preference" triggered by recent events. Extensive experiments on three public real-world datasets demonstrate TDPM's significant superiority, achieving average improvements of up to 29.21% in HR@20 and 25.45% in NDCG@20 over state-of-the-art baselines. An ablation study confirms the necessity of this time-aware token diffusion.

Key takeaway

For AI Scientists and Machine Learning Engineers designing generative recommendation systems, you should recognize that uniform diffusion processes overlook crucial time-evolving user preferences. TDPM offers a superior paradigm by explicitly integrating time-aware diffusion and disentangling preferences into period and point components. Consider adopting this approach to significantly enhance recommendation accuracy, as demonstrated by its substantial improvements in HR@20 and NDCG@20.

Key insights

TDPM enhances generative recommenders by integrating time-aware diffusion and disentangling user preferences into period and point components.

Principles

• User preference is non-stationary.
• Disentangle long-term and short-term preferences.
• Time-aware diffusion improves GRs.

Method

TDPM explicitly integrates time-evolving user preferences into the diffusion process on SID tokens by disentangling them into consistent "period preference" and event-triggered "point preference."

In practice

• Model user preferences as time-evolving.
• Incorporate temporal dynamics into diffusion.
• Separate stable from focal preferences.

Topics

• Generative Recommenders
• Diffusion Models
• Time-Aware Systems
• Preference Disentanglement
• Semantic Indices
• Information Retrieval

Best for: Research Scientist, AI Scientist, Machine Learning Engineer

Related on AIssential

• Case study: AI-assisted animation let a solo creator produce a 17-minute anime pilot — AI-assisted animation empowers solo creators to produce complex projects previously requiring large teams.
• Ad Creative Is A Technology Problem And Opportunity — Creative adtech is maturing, shifting from productivity to performance and integrated systems, driven by AI.
• Best Sudowrite Alternatives in 2026: 8 AI Writing Tools I Tested on Real Fiction Projects — AI writing tools for fiction offer creative support, but complex pricing models like Sudowrite's can hinder adoption.
• We’ve Reached the Point Where AI Has to Catch AI (The Internet Is Full of Slop) — AI detection tools are crucial for verifying content authenticity amidst widespread AI-generated text.
• Drifting Preference Optimization for One-Step Generative Models — DrPO enables efficient preference finetuning for one-step generative models using black-box rewards and feature-space updates.
• Google AI Studio vs Gemini App: What’s the Difference? — Gemini App is for using AI, while Google AI Studio is for building and prototyping with Gemini models.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.