[Paper Review] A Survey on Session-based Recommender Systems

Shoujin Wang, Longbing Cao+4 more|arXiv (Cornell University)|Feb 13, 2019

Recommender Systems and TechniquesComputer Science217 references76 citations

TL;DR

This paper provides a comprehensive, unified survey of Session-based Recommender Systems (SBRSs), introducing a unified problem statement, a taxonomy, data characteristics, challenges, and open research directions.

ABSTRACT

Recommender systems (RSs) have been playing an increasingly important role for informed consumption, services, and decision-making in the overloaded information era and digitized economy. In recent years, session-based recommender systems (SBRSs) have emerged as a new paradigm of RSs. Different from other RSs such as content-based RSs and collaborative filtering-based RSs which usually model long-term yet static user preferences, SBRSs aim to capture short-term but dynamic user preferences to provide more timely and accurate recommendations sensitive to the evolution of their session contexts. Although SBRSs have been intensively studied, neither unified problem statements for SBRSs nor in-depth elaboration of SBRS characteristics and challenges are available. It is also unclear to what extent SBRS challenges have been addressed and what the overall research landscape of SBRSs is. This comprehensive review of SBRSs addresses the above aspects by exploring in depth the SBRS entities (e.g., sessions), behaviours (e.g., users' clicks on items) and their properties (e.g., session length). We propose a general problem statement of SBRSs, summarize the diversified data characteristics and challenges of SBRSs, and define a taxonomy to categorize the representative SBRS research. Finally, we discuss new research opportunities in this exciting and vibrant area.

Motivation & Objective

• Define a unified SBRS problem statement built on core concepts (user, item, action, interaction, session).
• Characterize session data properties and the unique challenges SBRSs face.
• Provide a taxonomy and systematic comparison of representative SBRS approaches.
• Clarify SBRSs versus sequence-aware recommender systems (SRSs) and resolve inconsistencies in the literature.
• Identify open issues and promising directions for SBRS research.

Proposed method

• Propose a unified framework to categorize SBRS studies and a formal SBRS problem definition.
• Summarize diversified data characteristics and challenges arising from session data.
• Classify representative SBRS approaches into a taxonomy with brief technical details.
• Differentiate SBRSs from SRSs and outline the research landscape and gaps.
• Discuss open issues and future prospects to guide ongoing SBRS research.

Experimental results

Research questions

• RQ1 What are the core entities, behaviors, and properties that define SBRSs?
• RQ2 How can SBRS research be unified through a formal problem statement and framework?
• RQ3 What are the main characteristics and challenges of session data impacting SBRS design?
• RQ4 How are SBRS approaches currently classified, and what progress has been made across categories?
• RQ5 What open issues and future directions best guide SBRS research?

Key findings

• A unified SBRS framework reduces ambiguity and inconsistencies in the field.
• A formal problem statement ties SBRS to core concepts like session context and utility-based prediction.
• A comprehensive overview of session data characteristics and challenges is provided for SBRS design.
• A systematic classification and comparison of SBRS approaches is presented to map progress.
• The paper discusses open issues and future opportunities to guide SBRS research.