# A Survey on Session-based Recommender Systems

**Authors:** Shoujin Wang, Longbing Cao, Yan Wang, Quan Z. Sheng, Mehmet Orgun, Defu Lian  
**arXiv:** [1902.04864](https://arxiv.org/abs/1902.04864)  
**Date:** 2019-02-13  
**Summary page:** https://www.summarizepaper.com/en/paper/1902.04864v3/

## Key points

- Recommender systems (RSs) are essential tools in today's digital economy
- Session-based recommender systems (SBRSs) focus on capturing short-term and dynamic user preferences
- SBRSs offer timely and accurate recommendations that adapt to evolving session contexts
- Lack of unified problem statements and comprehensive exploration of SBRS characteristics and challenges
- Entities within SBRSs include sessions, user behaviors like clicks on items, and properties like session length
- Survey proposes a general problem statement for SBRSs and highlights diverse data characteristics and challenges
- Taxonomy is defined to categorize representative SBRS research, providing valuable insights into the current research landscape
- New research opportunities are identified in the field of session-based recommender systems

## Summary

Recommender systems (RSs) have become essential tools in today's information-rich and digital economy. They aid users in making informed decisions and enhancing their consumption experiences. Among the various types of RSs, session-based recommender systems (SBRSs) have emerged as a novel paradigm that focuses on capturing short-term and dynamic user preferences. These systems offer timely and accurate recommendations that adapt to evolving session contexts. Despite the increasing interest in SBRSs, there is a lack of unified problem statements and comprehensive exploration of their characteristics and challenges. The authors delve into the entities within SBRSs such as sessions, user behaviors like clicks on items, and properties like session length. The survey not only proposes a general problem statement for SBRSs but also highlights the diverse data characteristics and challenges faced by these systems. By defining a taxonomy to categorize representative SBRS research, the authors provide valuable insights into the current research landscape of SBRSs. Additionally, they identify new research opportunities in this dynamic field. This comprehensive review serves as a cornerstone for understanding the intricacies of SBRSs and sheds light on potential avenues for further exploration. With 39 pages, 163 references, 11 sections, 4 figures, and 11 tables, this survey stands out as the latest and most comprehensive examination of session-based recommender systems. Accepted by ACM Computing Surveys, it offers a deep dive into an exciting and vibrant area of research with significant implications for personalized recommendation services.