Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Survey on Sequential Recommendation

Liwei Pan, Weike Pan, Meiyan Wei, Hongzhi Yin, Zhong Ming

TL;DR

The paper surveys sequential recommendation through a property-centric lens, classifying methods into pure ID-based SR, SR with side information, and recent advancements including multi-modal, generative, and LLM-powered approaches. It highlights how item properties (IDs, general features, multi-modal content) and graph/review signals shape model design, transferability, and data efficiency, informing both traditional and modern architectures. Key contributions include a comprehensive taxonomy, synthesis of empirical findings across datasets, and forward-looking directions such as open-domain SR, data-centric strategies, cloud-edge collaboration, continuous adaptation, and explainable SR. The work emphasizes the practical implications of combining item IDs with side information and semantic signals to address cold-start, ultra-long sequences, and data sparsity, while outlining paths for responsible and scalable deployment.

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.

A Survey on Sequential Recommendation

TL;DR

The paper surveys sequential recommendation through a property-centric lens, classifying methods into pure ID-based SR, SR with side information, and recent advancements including multi-modal, generative, and LLM-powered approaches. It highlights how item properties (IDs, general features, multi-modal content) and graph/review signals shape model design, transferability, and data efficiency, informing both traditional and modern architectures. Key contributions include a comprehensive taxonomy, synthesis of empirical findings across datasets, and forward-looking directions such as open-domain SR, data-centric strategies, cloud-edge collaboration, continuous adaptation, and explainable SR. The work emphasizes the practical implications of combining item IDs with side information and semantic signals to address cold-start, ultra-long sequences, and data sparsity, while outlining paths for responsible and scalable deployment.

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.

Paper Structure

This paper contains 62 sections, 27 equations, 13 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Related Surveys
  3. Sequential Recommendation
  4. Problem Definition
  5. Properties
  6. Categories
  7. Taxonomy
  8. Challenges
  9. Pure ID-based SR
  10. Traditional Models
  11. Deep Learning Models
  12. Recurrent Neural Networks
  13. Convolutional Neural Networks
  14. Transformer
  15. Diffusion Model
  16. ...and 47 more sections

Figures (13)

  • Figure 1: The number of articles on sequential recommendation published between 2015 and 2024.
  • Figure 2: Properties in sequential recommendation.
  • Figure 3: The conversion between multi-modal features and general features.
  • Figure 4: An illustration of sequential recommendation.
  • Figure 5: Taxonomy of existing works on sequential recommendation.
  • ...and 8 more figures