Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PyGraft: Configurable Generation of Synthetic Schemas and Knowledge Graphs at Your Fingertips

Nicolas Hubert, Pierre Monnin, Mathieu d'Aquin, Davy Monticolo, Armelle Brun

TL;DR

The paper tackles the scarcity and bias of public KGs and the need for domain-agnostic benchmarks by introducing PyGraft, a Python tool that can generate both customizable schemas and knowledge graphs in a single pipeline and verify their logical consistency using a description logic reasoner. The method combines a schema generator (class and relation builders) with a KG generator that assigns entities to classes and creates triples, all while enforcing OWL/RDFS constraints and consistency checks. Its key contributions include domain-agnostic schema and KG generation, an integrated pipeline with DL reasoning, and demonstrated scalability across configurations (e.g., up to 100K entities and 1M triples with first-pass consistency). The tool enables diverse benchmarking for graph-based learning and KG processing, supports privacy-preserving data generation in sensitive domains, and fosters schema-driven, neuro-symbolic approaches, with an open-source release to encourage community contributions.

Abstract

Knowledge graphs (KGs) have emerged as a prominent data representation and management paradigm. Being usually underpinned by a schema (e.g., an ontology), KGs capture not only factual information but also contextual knowledge. In some tasks, a few KGs established themselves as standard benchmarks. However, recent works outline that relying on a limited collection of datasets is not sufficient to assess the generalization capability of an approach. In some data-sensitive fields such as education or medicine, access to public datasets is even more limited. To remedy the aforementioned issues, we release PyGraft, a Python-based tool that generates highly customized, domain-agnostic schemas and KGs. The synthesized schemas encompass various RDFS and OWL constructs, while the synthesized KGs emulate the characteristics and scale of real-world KGs. Logical consistency of the generated resources is ultimately ensured by running a description logic (DL) reasoner. By providing a way of generating both a schema and KG in a single pipeline, PyGraft's aim is to empower the generation of a more diverse array of KGs for benchmarking novel approaches in areas such as graph-based machine learning (ML), or more generally KG processing. In graph-based ML in particular, this should foster a more holistic evaluation of model performance and generalization capability, thereby going beyond the limited collection of available benchmarks. PyGraft is available at: https://github.com/nicolas-hbt/pygraft.

PyGraft: Configurable Generation of Synthetic Schemas and Knowledge Graphs at Your Fingertips

TL;DR

The paper tackles the scarcity and bias of public KGs and the need for domain-agnostic benchmarks by introducing PyGraft, a Python tool that can generate both customizable schemas and knowledge graphs in a single pipeline and verify their logical consistency using a description logic reasoner. The method combines a schema generator (class and relation builders) with a KG generator that assigns entities to classes and creates triples, all while enforcing OWL/RDFS constraints and consistency checks. Its key contributions include domain-agnostic schema and KG generation, an integrated pipeline with DL reasoning, and demonstrated scalability across configurations (e.g., up to 100K entities and 1M triples with first-pass consistency). The tool enables diverse benchmarking for graph-based learning and KG processing, supports privacy-preserving data generation in sensitive domains, and fosters schema-driven, neuro-symbolic approaches, with an open-source release to encourage community contributions.

Abstract

Knowledge graphs (KGs) have emerged as a prominent data representation and management paradigm. Being usually underpinned by a schema (e.g., an ontology), KGs capture not only factual information but also contextual knowledge. In some tasks, a few KGs established themselves as standard benchmarks. However, recent works outline that relying on a limited collection of datasets is not sufficient to assess the generalization capability of an approach. In some data-sensitive fields such as education or medicine, access to public datasets is even more limited. To remedy the aforementioned issues, we release PyGraft, a Python-based tool that generates highly customized, domain-agnostic schemas and KGs. The synthesized schemas encompass various RDFS and OWL constructs, while the synthesized KGs emulate the characteristics and scale of real-world KGs. Logical consistency of the generated resources is ultimately ensured by running a description logic (DL) reasoner. By providing a way of generating both a schema and KG in a single pipeline, PyGraft's aim is to empower the generation of a more diverse array of KGs for benchmarking novel approaches in areas such as graph-based machine learning (ML), or more generally KG processing. In graph-based ML in particular, this should foster a more holistic evaluation of model performance and generalization capability, thereby going beyond the limited collection of available benchmarks. PyGraft is available at: https://github.com/nicolas-hbt/pygraft.
Paper Structure (14 sections, 4 figures, 5 tables, 3 algorithms)

This paper contains 14 sections, 4 figures, 5 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. PyGraft Description
  4. Preliminaries
  5. Overview
  6. Schema Generation
  7. Knowledge Graph Generation
  8. PyGraft in Action
  9. Efficiency and Scalability Details
  10. Usage Illustration
  11. Discussion
  12. Potential Uses
  13. Limitations, Sustainability, Maintenance and Future Work
  14. Conclusion

Figures (4)

  • Figure 1: PyGraft general overview.
  • Figure 2: Potential class hierarchies for the constraints: $\texttt{num\_classes} = 6$, $\texttt{max\_depth} = 3$, $\texttt{avg\_depth} = 1.5$, and $\texttt{inheritance\_ratio} = 2.5$. Left and middle class hierarchies are built with parameter priority. The right class hierarchy is built with a best-effort strategy, without specific parameter privilege.
  • Figure 3: Execution time breakdown for each configuration.
  • Figure : Class Generation