Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

RAIE: Region-Aware Incremental Preference Editing with LoRA for LLM-based Recommendation

Jin Zeng, Yupeng Qi, Hui Li, Chengming Li, Ziyu Lyu, Lixin Cui, Lu Bai

TL;DR

Experiments show that RAIE significantly outperforms state-of-the-art baselines while effectively mitigating forgetting, and demonstrate that region-aware editing offers an accurate and scalable mechanism for continual adaptation in dynamic recommendation scenarios.

Abstract

Large language models (LLMs) are increasingly adopted as the backbone of recommender systems. However, user-item interactions in real-world scenarios are non-stationary, making preference drift over time inevitable. Existing model update strategies mainly rely on global fine-tuning or pointwise editing, but they face two fundamental challenges: (i) imbalanced update granularity, where global updates perturb behaviors unrelated to the target while pointwise edits fail to capture broader preference shifts; (ii) unstable incremental updates, where repeated edits interfere with prior adaptations, leading to catastrophic forgetting and inconsistent recommendations. To address these issues, we propose Region-Aware Incremental Editing (RAIE), a plug-in framework that freezes the backbone model and performs region-level updates. RAIE first constructs semantically coherent preference regions via spherical k-means in the representation space. It then assigns incoming sequences to regions via confidence-aware gating and performs three localized edit operations - Update, Expand, and Add - to dynamically revise the affected region. Each region is equipped with a dedicated Low-Rank Adaptation (LoRA) module, which is trained only on the region's updated data. During inference, RAIE routes each user sequence to its corresponding region and activates the region-specific adapter for prediction. Experiments on two benchmark datasets under a time-sliced protocol that segments data into Set-up (S), Finetune (F), and Test (T) show that RAIE significantly outperforms state-of-the-art baselines while effectively mitigating forgetting. These results demonstrate that region-aware editing offers an accurate and scalable mechanism for continual adaptation in dynamic recommendation scenarios. Our code is available at https://github.com/fengaogao/RAIE.

RAIE: Region-Aware Incremental Preference Editing with LoRA for LLM-based Recommendation

TL;DR

Experiments show that RAIE significantly outperforms state-of-the-art baselines while effectively mitigating forgetting, and demonstrate that region-aware editing offers an accurate and scalable mechanism for continual adaptation in dynamic recommendation scenarios.

Abstract

Large language models (LLMs) are increasingly adopted as the backbone of recommender systems. However, user-item interactions in real-world scenarios are non-stationary, making preference drift over time inevitable. Existing model update strategies mainly rely on global fine-tuning or pointwise editing, but they face two fundamental challenges: (i) imbalanced update granularity, where global updates perturb behaviors unrelated to the target while pointwise edits fail to capture broader preference shifts; (ii) unstable incremental updates, where repeated edits interfere with prior adaptations, leading to catastrophic forgetting and inconsistent recommendations. To address these issues, we propose Region-Aware Incremental Editing (RAIE), a plug-in framework that freezes the backbone model and performs region-level updates. RAIE first constructs semantically coherent preference regions via spherical k-means in the representation space. It then assigns incoming sequences to regions via confidence-aware gating and performs three localized edit operations - Update, Expand, and Add - to dynamically revise the affected region. Each region is equipped with a dedicated Low-Rank Adaptation (LoRA) module, which is trained only on the region's updated data. During inference, RAIE routes each user sequence to its corresponding region and activates the region-specific adapter for prediction. Experiments on two benchmark datasets under a time-sliced protocol that segments data into Set-up (S), Finetune (F), and Test (T) show that RAIE significantly outperforms state-of-the-art baselines while effectively mitigating forgetting. These results demonstrate that region-aware editing offers an accurate and scalable mechanism for continual adaptation in dynamic recommendation scenarios. Our code is available at https://github.com/fengaogao/RAIE.
Paper Structure (30 sections, 8 equations, 5 figures, 2 tables, 1 algorithm)

This paper contains 30 sections, 8 equations, 5 figures, 2 tables, 1 algorithm.

Table of Contents

  1. INTRODUCTION
  2. RELATED WORK
  3. Continual learning
  4. Parameter-Efficient Fine-Tuning for LLM-based Recommendation systems
  5. Knowledge Editing for LLMs and Recommenders
  6. Problem Definition
  7. Set-up Phase (S)
  8. Incremental Learning Finetune Phase (F)
  9. Inference Test Phase (T)
  10. Method
  11. Knowledge Region Construction
  12. Subsequence Segmentation and representation
  13. Preference Region Discovery
  14. Regional LoRA Mapping
  15. Region-Aware Preference Editing and LoRA Adaptation
  16. ...and 15 more sections

Figures (5)

  • Figure 1: Illustrated example of user preference drift in the movie domain.
  • Figure 2: The overall architecture of our proposed RAIE. It consists of three core modules, respectively. a) Knowledge Region Construction at the set-up phase (S); b) Region-aware Editing and LoRA Adaption in the incremental learning finetune phase (F); and the Region-ware Routing for the inference test phase (T).
  • Figure 3: The ablation study of RAIE.
  • Figure 4: Impact of region number K.
  • Figure 5: The visualization of region distributions.