Research

My research focuses on making materials science data FAIR (Findable, Accessible, Interoperable, Reusable) through knowledge graphs, ontologies, and semantic technologies. I work at the intersection of Knowledge Engineering, Materials Informatics, and AI.

The Big Picture

Materials science generates vast amounts of heterogeneous data — from lab experiments and simulations to published literature. But this data is often trapped in silos: proprietary databases, PDFs, unstructured spreadsheets, and disconnected repositories. My work aims to bridge these silos by creating a shared semantic layer that machines and humans can both understand.

MSE Knowledge Graph (MSE-KG)

The MSE-KG connects researchers, organizations, datasets, publications, and experimental workflows into a single queryable graph. It is built on top of the NFDI-MatWerk Ontology (MWO) and the NFDIcore ontology framework.

Key contributions:

• Centralized metadata management for the NFDI-MatWerk consortium
• SPARQL endpoint for querying across heterogeneous materials science data
• Integration with external sources (Wikidata, ORCID, ROR, DataCite)
• Won “Best Demo” at NFDI-MatWerk AHoD 2026

Ontology Design Patterns for Materials Science

Ontologies in materials science are often complex and hard to reuse. Ontology Design Patterns (ODPs) offer modular, reusable solutions — like design patterns in software engineering, but for knowledge modeling.

My work surveys existing ontologies, extracts implicit design patterns from their structures, and proposes standardized patterns for common modeling problems in MSE:

• Process and workflow representation
• Material composition and structure
• Measurement and characterization
• Provenance and experimental metadata

Key contributions:

• Comprehensive survey of MSE ontologies and their process modeling capabilities
• Baseline method for automatic ODP extraction from existing ontologies
• Open-source pattern catalog: ODPs4MSE on GitHub

NFDI-MatWerk Ontology (MWO)

The MWO is a BFO-compliant domain ontology I co-developed for research data management in materials science. It serves as the backbone for the MSE-KG and is designed to:

• Align with the Basic Formal Ontology (BFO) upper-level framework
• Provide domain-specific classes for materials, processes, and properties
• Enable interoperability across NFDI consortia (MatWerk, Culture, DS, Memory, Chem)

Chemotion Knowledge Graph

As part of the AI4DiTraRe project, I built a BFO-compliant knowledge graph from experimental chemistry data in the Chemotion electronic lab notebook system.

This KG enables:

• Cross-referencing experimental procedures with ontological concepts
• Provenance tracking from raw measurements to published results
• AI-driven discovery of related experiments and materials

Chemotion-KG

LLMs for Knowledge Engineering

I’m also exploring how Large Language Models can assist with:

• Ontology alignment and matching (OAEI campaigns)
• Information extraction from scientific literature
• Automated knowledge graph construction
• Concept extraction using NLP pipelines (ConExion)

Collaborations & Projects

See my full publication list or explore my GitHub projects for code and datasets.