Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

RecKG: Knowledge Graph for Recommender Systems

Junhyuk Kwon, Seokho Ahn, Young-Duk Seo

TL;DR

RecKG tackles the interoperability gap among knowledge-graph–based recommender systems by introducing a standardized KG focused on consistent user/item attributes and interactions. The authors define design principles that foreground consistency and diversity, instantiate RecKG on real-world datasets (e.g., MovieLens and Yahoo!Movies), and implement a graph-database application to demonstrate cross-dataset interoperability. Through qualitative comparisons with existing KG-based recommender systems, RecKG shows enhanced semantic integration and supports higher-order, cross-domain reasoning, enabling richer explanations and recommendations. This work lays a practical foundation for cross-domain KG integration in recommender research and points to future expansion across more datasets and domains.

Abstract

Knowledge graphs have proven successful in integrating heterogeneous data across various domains. However, there remains a noticeable dearth of research on their seamless integration among heterogeneous recommender systems, despite knowledge graph-based recommender systems garnering extensive research attention. This study aims to fill this gap by proposing RecKG, a standardized knowledge graph for recommender systems. RecKG ensures the consistent representation of entities across different datasets, accommodating diverse attribute types for effective data integration. Through a meticulous examination of various recommender system datasets, we select attributes for RecKG, ensuring standardized formatting through consistent naming conventions. By these characteristics, RecKG can seamlessly integrate heterogeneous data sources, enabling the discovery of additional semantic information within the integrated knowledge graph. We apply RecKG to standardize real-world datasets, subsequently developing an application for RecKG using a graph database. Finally, we validate RecKG's achievement in interoperability through a qualitative evaluation between RecKG and other studies.

RecKG: Knowledge Graph for Recommender Systems

TL;DR

RecKG tackles the interoperability gap among knowledge-graph–based recommender systems by introducing a standardized KG focused on consistent user/item attributes and interactions. The authors define design principles that foreground consistency and diversity, instantiate RecKG on real-world datasets (e.g., MovieLens and Yahoo!Movies), and implement a graph-database application to demonstrate cross-dataset interoperability. Through qualitative comparisons with existing KG-based recommender systems, RecKG shows enhanced semantic integration and supports higher-order, cross-domain reasoning, enabling richer explanations and recommendations. This work lays a practical foundation for cross-domain KG integration in recommender research and points to future expansion across more datasets and domains.

Abstract

Knowledge graphs have proven successful in integrating heterogeneous data across various domains. However, there remains a noticeable dearth of research on their seamless integration among heterogeneous recommender systems, despite knowledge graph-based recommender systems garnering extensive research attention. This study aims to fill this gap by proposing RecKG, a standardized knowledge graph for recommender systems. RecKG ensures the consistent representation of entities across different datasets, accommodating diverse attribute types for effective data integration. Through a meticulous examination of various recommender system datasets, we select attributes for RecKG, ensuring standardized formatting through consistent naming conventions. By these characteristics, RecKG can seamlessly integrate heterogeneous data sources, enabling the discovery of additional semantic information within the integrated knowledge graph. We apply RecKG to standardize real-world datasets, subsequently developing an application for RecKG using a graph database. Finally, we validate RecKG's achievement in interoperability through a qualitative evaluation between RecKG and other studies.
Paper Structure (18 sections, 4 figures, 2 tables)

This paper contains 18 sections, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Knowledge Graph Integration Across Diverse Fields
  4. Knowledge Graph-based Recommender System
  5. Knowledge Graph for Recommender Systems
  6. Goal of RecKG
  7. Attributes of RecKG
  8. Attributes of Users
  9. Attributes of Items
  10. User-Item Interactions
  11. Design based on RecKG
  12. Case Study and Evaluation
  13. Integration of Real-World Dataset
  14. Instance of RecKG
  15. Integration of RecKG
  16. ...and 3 more sections

Figures (4)

  • Figure 1: The diagram of RecKG. Nodes represent user and item attributes, with gray nodes indicating text or image attributes. Edges denote relations between nodes.
  • Figure 2: Design based on RecKG. (a) and (b) show KG designs for the MovieLens and Yahoo!Movies datasets using RecKG, respectively. (c) illustrates the integrated KG design, obtained by combining the KG designs presented in (b) into (a).
  • Figure 3: Instances of RecKG. (a) and (b) show sample subgraphs from the MovieLens and Yahoo!Movies datasets in RecKG. (c) illustrates a sample subgraph when integrating Yahoo!Movies into MovieLens in RecKG.
  • Figure 4: Integrating Yahoo!Movies into MovieLens on RecKG