AdaRec: Adaptive Recommendation with LLMs via Narrative Profiling and Dual-Channel Reasoning

2025-11-10

AI Summary

Existing recommender systems suffer from heavy reliance on manual feature engineering, poor cross-task transferability, and limited effectiveness in few-shot, zero-shot, and long-tail scenarios. To address these challenges, this paper proposes AdaRec—a large language model (LLM)-based adaptive personalized recommendation framework. Its core contributions are threefold: (1) narrative user profiling, which converts raw interaction sequences into natural-language behavioral descriptions; (2) a dual-channel reasoning mechanism that jointly performs horizontal behavioral alignment and vertical causal attribution, enabling interpretable, feature-engineering-free unified modeling; and (3) a lightweight adaptation strategy integrating synthetic data generation with parameter-efficient fine-tuning to facilitate rapid cross-task transfer. Evaluated on real-world e-commerce datasets, AdaRec achieves up to 8% improvement over state-of-the-art methods in few-shot settings and outperforms expert-crafted user profiles by 19% in zero-shot settings—while attaining full-parameter fine-tuning performance with only lightweight adaptation.

Technology Category

• Data Mining & Knowledge Management: Recommender Systems
• Cognitive Modeling & Cognitive Systems: Adaptive Behavior
• Machine Learning: Large Multimodal Models (LMMs)

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
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

We propose AdaRec, a few-shot in-context learning framework that leverages large language models for an adaptive personalized recommendation. AdaRec introduces narrative profiling, transforming user-item interactions into natural language representations to enable unified task handling and enhance human readability. Centered on a bivariate reasoning paradigm, AdaRec employs a dual-channel architecture that integrates horizontal behavioral alignment, discovering peer-driven patterns, with vertical causal attribution, highlighting decisive factors behind user preferences. Unlike existing LLM-based approaches, AdaRec eliminates manual feature engineering through semantic representations and supports rapid cross-task adaptation with minimal supervision. Experiments on real ecommerce datasets demonstrate that AdaRec outperforms both machine learning models and LLM-based baselines by up to eight percent in few-shot settings. In zero-shot scenarios, it achieves up to a nineteen percent improvement over expert-crafted profiling, showing effectiveness for long-tail personalization with minimal interaction data. Furthermore, lightweight fine-tuning on synthetic data generated by AdaRec matches the performance of fully fine-tuned models, highlighting its efficiency and generalization across diverse tasks.

Problem

• Adaptive personalized recommendation using LLMs with minimal supervision
• Transforming user-item interactions into natural language narrative profiles
• Integrating behavioral alignment and causal attribution for preference reasoning

Innovation

• Narrative profiling transforms interactions into natural language
• Dual-channel architecture integrates behavioral alignment with causal attribution
• Eliminates manual feature engineering through semantic representations