A Survey on Sequential Recommendation

2024-12-17

arXiv.org

AI Summary

This paper addresses sequential recommendation—modeling temporal dependencies in user interaction sequences to capture dynamic preferences accurately. We propose a unified taxonomy centered on “item attribute construction,” systematically surveying technical advances from 2018 to 2024 across six paradigms: ID-driven modeling, multimodal fusion, generative modeling, large language model (LLM) augmentation, ultra-long sequence processing, and data augmentation. We further identify seven emerging research directions: open-domain recommendation, data-centric recommendation, edge-cloud collaboration, continual learning, ethical (beneficence-oriented) recommendation, and explainability. To consolidate this landscape, we construct the first end-to-end knowledge map for sequential recommendation, integrating key techniques—including graph neural networks, LLM prompting, diffusion models, and long-context modeling. This map serves as an authoritative roadmap bridging academic research and industrial deployment.

Technology Category

• Data Mining & Knowledge Management: Recommender Systems
• Reasoning under Uncertainty: Sequential Decision Making
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Different from most conventional recommendation problems, sequential recommendation focuses on learning users' preferences by exploiting the internal order and dependency among the interacted items, which has received significant attention from both researchers and practitioners. In recent years, we have witnessed great progress and achievements in this field, necessitating a new survey. In this survey, we study the SR problem from a new perspective (i.e., the construction of an item's properties), and summarize the most recent techniques used in sequential recommendation such as pure ID-based SR, SR with side information, multi-modal SR, generative SR, LLM-powered SR, ultra-long SR and data-augmented SR. Moreover, we introduce some frontier research topics in sequential recommendation, e.g., open-domain SR, data-centric SR, could-edge collaborative SR, continuous SR, SR for good, and explainable SR. We believe that our survey could be served as a valuable roadmap for readers in this field.

Problem

• Explores sequential recommendation by analyzing item order and dependencies.
• Summarizes recent techniques like ID-based, multi-modal, and LLM-powered SR.
• Introduces frontier topics including open-domain and explainable sequential recommendation.

Innovation

• Exploits item order and dependency
• Utilizes multi-modal and generative techniques
• Incorporates LLM-powered and data-augmented methods