Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SUBMASSIVE: Resolving Subclass Cycles in Very Large Knowledge Graphs

Shuai Wang, Peter Bloem, Joe Raad, Frank van Harmelen

TL;DR

This work tackles erroneous cyclic subclass relations in very large knowledge graphs, which hinder reliable subclass hierarchies and transitive closures. It introduces SUBMASSIVE, a hybrid approach that combines data pre-processing with an iterative, local cycle-resolving procedure implemented via a MAXSAT formulation to remove a minimal set of rdfs:subClassOf edges. Evaluated on the LOD-a-lot dataset, the method demonstrates scalability and reveals a trade-off between the number of removed relations and computation time, achieving substantial cycle reduction with manageable runtime under controlled bounds. The resulting cycle-free hierarchy enables accurate transitive reasoning and can support downstream machine learning tasks on large-scale knowledge graphs. This work advances KG refinement by providing an anytime, scalable cycle-resolving method and releasing processed data for reproducibility.

Abstract

Large knowledge graphs capture information of a large number of entities and their relations. Among the many relations they capture, class subsumption assertions are usually present and expressed using the \texttt{rdfs:subClassOf} construct. From our examination, publicly available knowledge graphs contain many potentially erroneous cyclic subclass relations, a problem that can be exacerbated when different knowledge graphs are integrated as Linked Open Data. In this paper, we present an automatic approach for resolving such cycles at scale using automated reasoning by encoding the problem of cycle-resolving to a MAXSAT solver. The approach is tested on the LOD-a-lot dataset, and compared against a semi-automatic version of our algorithm. We show how the number of removed triples is a trade-off against the efficiency of the algorithm.

SUBMASSIVE: Resolving Subclass Cycles in Very Large Knowledge Graphs

TL;DR

This work tackles erroneous cyclic subclass relations in very large knowledge graphs, which hinder reliable subclass hierarchies and transitive closures. It introduces SUBMASSIVE, a hybrid approach that combines data pre-processing with an iterative, local cycle-resolving procedure implemented via a MAXSAT formulation to remove a minimal set of rdfs:subClassOf edges. Evaluated on the LOD-a-lot dataset, the method demonstrates scalability and reveals a trade-off between the number of removed relations and computation time, achieving substantial cycle reduction with manageable runtime under controlled bounds. The resulting cycle-free hierarchy enables accurate transitive reasoning and can support downstream machine learning tasks on large-scale knowledge graphs. This work advances KG refinement by providing an anytime, scalable cycle-resolving method and releasing processed data for reproducibility.

Abstract

Large knowledge graphs capture information of a large number of entities and their relations. Among the many relations they capture, class subsumption assertions are usually present and expressed using the \texttt{rdfs:subClassOf} construct. From our examination, publicly available knowledge graphs contain many potentially erroneous cyclic subclass relations, a problem that can be exacerbated when different knowledge graphs are integrated as Linked Open Data. In this paper, we present an automatic approach for resolving such cycles at scale using automated reasoning by encoding the problem of cycle-resolving to a MAXSAT solver. The approach is tested on the LOD-a-lot dataset, and compared against a semi-automatic version of our algorithm. We show how the number of removed triples is a trade-off against the efficiency of the algorithm.

Paper Structure

This paper contains 15 sections, 3 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Related Work
  4. Algorithm
  5. Data Pre-processing
  6. Automated Cycle-resolving
  7. Retrieve Local Simple Cycles.
  8. Resolving Simple Cycles Using MAXSAT.
  9. Implementation
  10. Evaluation
  11. Evaluation of the Fully Automated Version
  12. Evaluation against the Semi-automated Version
  13. Discussion
  14. Conclusion and Future Work
  15. Acknowledgements

Figures (3)

  • Figure 1: An example graph
  • Figure 2: Number of removed edges against the soft bound on size of subgraph at each iteration (error-bars represent the standard deviation).
  • Figure 3: An analysis of removed relations in stacked bar-chart.