Ontology-based knowledge graph infrastructure for interoperable atomistic simulation data

Abril Azocar Guzman; Sarath Menon; Tilmann Hickel; Stefan Sandfeld

cs.DBMar 31, 2026

Ontology-based knowledge graph infrastructure for interoperable atomistic simulation data

Authors: Abril Azocar Guzman, Sarath Menon, Tilmann Hickel, Stefan Sandfeld

Organizations: Institute for Advanced Simulations – Materials Data Science and Informatics (IAS-9), Forschungszentrum Jülich GmbH, Jülich, Germany · Interdisciplinary Centre for Advanced Materials Simulation (ICAMS), Ruhr University Bochum, Bochum, Germany · Bundesanstalt für Materialforschung und -prüfung (BAM), Berlin, Germany

Abstract

The reuse of atomistic simulation data is often limited by heterogeneous formats, incomplete metadata, and a lack of standardized representations of workflows and provenance. Here we present an ontology-based infrastructure for representing and integrating atomistic simulation data as a knowledge graph. The approach combines domain ontologies with a software framework that enables data capture both from existing datasets and directly from simulation workflows at the point of generation. Heterogeneous data from multiple sources are normalized into a common, ontology-aligned representation, enabling consistent querying and analysis across datasets. We demonstrate these capabilities through the integration of grain boundary data, cross-dataset analysis of material properties, and extraction of derived thermodynamic quantities from existing simulations. In addition, workflows are represented in a machine-readable form, enabling both forward provenance tracking and partial reconstruction of computational procedures. The resulting knowledge graph contains over 750,000 triples describing nearly 8,000 computational samples. This work provides a practical framework for improving the findability, interoperability, and reuse of atomistic simulation data.

Figures & tables

Figure 4: Visualisation of the knowledge graph of atomistic simulation data. Nodes represent RDF resources such as simulation samples, crystal structures, chemical elements, calculated properties, interatomic potentials, and computational methods; coloured by semantic category (orange: Sample; blue: Structure; teal: Property; red: Calculation; green: Element; purple: Material; gold: Potential). Labels are shown only for the ten most-connected nodes. The visualized data is subsampled from the knowledge graph (13% of the full graph).

Figure 5: Semantic integration of heterogeneous grain boundary data in the knowledge graph. The left panel summarizes the currently available grain-boundary-related properties across elements, methods, and data sources. The right panel shows the result of a targeted query for the energy of $\Sigma$ 3 grain boundaries calculated with different methods (n is the total number of data records), illustrating how semantically aligned records can be retrieved across datasets to reveal existing coverage and gaps in the current data landscape.

Figure 6: Cross-dataset grain boundary properties queried from the knowledge graph. (a) Grain boundary energy ( $E_{\mathrm{GB}}$ ) versus misorientation angle for Cu along the $\langle 110\rangle$ , $\langle 001\rangle$ , and $\langle 111\rangle$ tilt axes. Continuous curves are from molecular dynamics, while DFT reference values are marked with crosses. Within $\langle 110\rangle$ , $\Sigma 3$ (solid) and $\Sigma 5$ (dashed) boundaries are highlighted, with further data shown as dots. (b) Correlation between DFT vacancy formation energy ( $E_{\mathrm{vf}}$ ) and DFT median $E_{\mathrm{GB}}$ for eight transition-metal elements across four $\Sigma$ values.

Figure 7: Volume per atom as a function of temperature for selected elemental systems retrieved from the knowledge graph. By querying structurally and methodologically comparable molecular dynamics data, the volumetric thermal expansion coefficient can be estimated from the slope of the resulting volume-temperature relation.

Figure 8: Machine-readable provenance representation for vacancy formation energy calculations, comparing workflows based on DFT and molecular statics. Although the underlying simulation methods differ, the graph representation reveals their shared workflow structure and makes both common and method-specific steps explicit.

KG Construction

Knowledge graph construction turns text, tables, images, and databases into structured entities and relationships that machines can query and reason over.

See all 195 papers

A strategic roadmap for an atomistic machine-learning ecosystem
Sep 30, 2026Jörg Behler, Michele Ceriotti, Cecilia Clementi +47Machine Learning Interatomic Potentials Scientific ML
An Ontology for Machine Learning Interatomic Potentials
Jul 25, 2026Daniel Hernández, Jong Hyun Jung, Yuji Ikeda +11Machine Learning Interatomic Potentials
Institute for Artificial Intelligence, University of Stuttgart, Germany · Institute for Materials Science, University of Stuttgart, Germany · Interdisciplinary Centre for Advanced Materials Simulation (ICAMS), Ruhr-Universität Bochum, Germany and Department for Computational Materials Design, Max-Planck-Institut for Sustainable Materials, Germany
The AnIML Ontology: Enabling Semantic Interoperability for Large-Scale Experimental Data in Interconnected Scientific Labs
Apr 2, 2026Wilf Morlidge, Elliott Watkiss-Leek, George Hannah +7
School of Computer Science & Informatics, University of Liverpool, UK · Unilever Plc. Materials Innovation Factory, University of Liverpool, UK

Sep 30, 2026J/K move · Enter open · S save

A strategic roadmap for an atomistic machine-learning ecosystem

Jörg Behler, Michele Ceriotti, Cecilia Clementi, Gábor Csányi+46

Machine Learning Interatomic Potentials Scientific ML Computational Materials Science

Open paper AskPDF arXiv

Data-driven machine learning (ML) techniques have become an essential tool in many domains of science. Their application to atomistic simulations of matter is particularly widespread and impactful. This success is due largely to the existence of a well-developed and established physics-based modeling framework, ranging from first-principles electronic-structure calculations to molecular dynamics and statistical sampling, into which ML was integrated naturally to reshape long-standing trade-offs between accuracy, efficiency, and scale. Nevertheless, this integration raises both conceptual and practical challenges, from choosing between data-centric and physics-based modeling approaches to adapting established software stacks to modern hardware accelerators and ML libraries. As the field evolves rapidly, fueled in part by widespread enthusiasm but also by tangible impact, it seems appropriate to take a moment to consider the current state of the art and open challenges, and reflect on what can be done to better coordinate efforts across the community. With this goal in mind, several members of this community met in Lausanne in January 2026 at CECAM to discuss algorithms, models, software and hardware infrastructure, and the most promising scientific applications that have become possible thanks to the use of artificial intelligence in atomic-scale simulations. This strategic roadmap paper summarizes the outcomes of these discussions, suggesting some long-term goals, and some concrete actions, to establish a healthy, sustainable and impactful atomistic ML ecosystem.

Abstract

KG Construction

Explore similar work

A strategic roadmap for an atomistic machine-learning ecosystem

An Ontology for Machine Learning Interatomic Potentials

The AnIML Ontology: Enabling Semantic Interoperability for Large-Scale Experimental Data in Interconnected Scientific Labs