AI4DiTraRe: Building the BFO-Compliant Chemotion Knowledge Graph
Abstract
Chemistry is an example of a discipline where the advancements of technology have led to multi-level and often tangled and tricky processes ongoing in the lab. The repeatedly complex workflows are combined with information from chemical structures, which are essential to understand the scientific process. An important tool for many chemists is Chemotion, which consists of an electronic lab notebook and a repository. This paper introduces a semantic pipeline for constructing the BFO-compliant Chemotion Knowledge Graph, providing an integrated, ontology-driven representation of chemical research data. The Chemotion-KG has been developed to adhere to the FAIR (Findable, Accessible, Interoperable, Reusable) principles and to support AI-driven discovery and reasoning in chemistry. Experimental metadata were harvested from the Chemotion API in JSON-LD format. The JSON-LD, as an RDF serialization, was ingested into the triple store and subsequently transformed into a Basic Formal Ontology-aligned graph through SPARQL CONSTRUCT queries. The source code and datasets are publicly available via GitHub. The Chemotion Knowledge Graph is hosted by FIZ Karlsruhe Information Service Engineering. Outcomes presented in this work were achieved within the Leibniz Science Campus “Digital Transformation of Research” (DiTraRe) and are part of an ongoing interdisciplinary collaboration.
Full text
AI4DiTraRe: Building the BFO-Compliant Chemotion Knowledge Graph Ebrahim Norouzi, Nicole Jung, Anna M. Jacyszyn, Jörg Waitelonis and Harald Sack FIZ Karlsruhe & KIT(Institute of Applied Informatics and Formal Description Methods and Institute of Biological and Chemical Systems) Sci-K 2025, 5th International Workshop on Scientific Knowledge: Representation, Discovery, and Assessment @ ISWC, Nov 2nd, 2025, Nara, Japan
Agenda •Motivation •Knowledge Graph Construction Approach •Ontology Integration, Why? •Ontology Design Patterns (ODPs), Why? •Results & Statistics •Limitations •Future Work & Summary
DiTraRe Leibniz Science Campus Digital Transformation of Research + ■Growth core to establish new research branch. ■Consortium: 4+4 years (start: September 2023). ■Multilevel approach. → We start with 4 interdisciplinary use cases. ■AI4DiTraRe: solutions based on different AI methods to solve research questions in numerous areas of research. Study overall influence and effects of AI. This work (FIZ-Karlsruhe & KIT Institute of Biological and Chemical Systems): Chemotion Electronic Lab Notebook + AI
4 Motivation •Increasing complexity of chemistry workflows, multimodal data (spectra, reactions, metadata) •Chemotion Repository: rich chemistry data (reactions, spectra, studies, datasets) ■Schema.org-based structure too shallow for reasoning or linking ■No semantic alignment with other domain ontologies ■Hard to connect/query across external resources Ebrahim Norouzi et al., Chemotion Knowledge Graph, Sci-K 2025 @ ISWC, Nov 2nd, Nara, Japan
5 Goal •Transform Chemotion repository into a machine-understandable Knowledge Graph (KG) •Enable: ■Reasoning: infer hidden relations (e.g., link identical chemicals across studies) ■Provenance tracking: trace how data evolved or who contributed ■Cross-KG integration: connect Chemotion with NFDI4Chem, PubChem, MatWerk, etc. ■Provide semantically enriched chemistry data for discovery and reuse Ebrahim Norouzi et al., Chemotion Knowledge Graph, Sci-K 2025 @ ISWC, Nov 2nd, Nara, Japan
6 Introduction: Why we need top-level ontologies? Enable integration across different domain ontologies Avoid ambiguity by defining core categories (e.g., object, process, quality) Logical structure helps automated reasoning and validation Allow systems and datasets to “speak the same language” Save time and effort by building on established upper-level concepts
7 Introduction: What is BFO? •Basic Formal Ontology (BFO) [1] → one of the leading upper-level ontology •Widely adopted, philosophically grounded in realism •Ensures interoperability & reusability across domains •Clear and consistent class hierarchy •Stable across versions → trust and adoption •Foundation for many domain ontologies [1] BFO guidebook: https://direct.mit.edu/books/monograph/4044/Building-Ontologies-with-Basic-Formal-Ontology
8 Knowledge Graph Construction Approach •Source: Chemotion Repository (schema.org structure): lacks upper-level semantics, Incompatible with external ontologies and reasoning •Target: BFO-compliant, ontology-aligned RDF knowledge graph •Objective: semantic enrichment → reasoning, integrate with external KGs, enable cross-domain queries •Ontology Design Patterns (ODPs) applied during schema-to-ontology mapping stage Ebrahim Norouzi et al., Chemotion Knowledge Graph, Sci-K 2025 @ ISWC, Nov 2nd, Nara, Japan
9 Ontology Integration & Selection Rationale •Upper-level alignment: ■BFO – Basic Formal Ontology (provides upper-level semantics for interoperability) ■Widely adopted in life sciences & research data infrastructures ■Enables logical entailment and cross-domain reasoning Ebrahim Norouzi et al., Chemotion Knowledge Graph, Sci-K 2025 @ ISWC, Nov 2nd, Nara, Japan •Middle-level alignment: ■NFDICore Ontology (provides modeling for representing metadata related to resources, including individuals, organizations, projects, data portals, and more) - Already BFO-aligned ■Acts as a bridge layer between abstract BFO classes and domain-specific ontologies. ■Developed by FIZ Karlsruhe for representing resources in the National Research Data Infrastructure (NFDI) initiatives & facilitates interoperability within NFDI ecosystems (e.g., NFDI4Culture, NFDIMatWerk) •Domain-level alignment: ■ChEBI – Chemical Entities of Biological Interest (provides controlled vocabulary for chemical substances and molecular entities) - Already BFO-aligned
16 Chemical Substance Pattern Ebrahim Norouzi et al., Chemotion Knowledge Graph, Sci-K 2025 @ ISWC, Nov 2nd, Nara, Japan
17 Integrated Graph View Ebrahim Norouzi et al., Chemotion Knowledge Graph, Sci-K 2025 @ ISWC, Nov 2nd, Nara, Japan
18 Knowledge Graph Statistics •Total triples: 1,462,187 •Total instances: 87,782 ■Datasets 20,701 ■Studies 20,563 ■Chemicals 3,746 ■Creators 250 ■Chemical substances (obo:CHEBI_59999): 4,923 instances •Public SPARQL endpoint: https://ditrare.ise.fiz-karlsruhe.de/chemotion-kg/sparql Ebrahim Norouzi et al., Chemotion Knowledge Graph, Sci-K 2025 @ ISWC, Nov 2nd, Nara, Japan
19 Limitations •Scope restriction ■Current KG models metadata layer only -> Raw instrument outputs (e.g., NMR, MS, IR spectral files) not yet semantically integrated •Canonical IRIs generated by concatenating Chemotion resource IDs (e.g. “https://ditrare.ise.fiz-karlsruhe.de/chemotion-kg/resources/2025-07/Dataset_1 e332f08-cd6d-41b9-a6b8”) are Long. Ebrahim Norouzi et al., Chemotion Knowledge Graph, Sci-K 2025 @ ISWC, Nov 2nd, Nara, Japan
20 Future Work & Summary •Extend KG → include instrument and measurement data •Cross-link to PubChem RDF, ChemSpider, NFDI4Chem •Develop SHACL validations •Evaluate user queries and domain relevance •Integrate LLMs semantic alignment and data enrichment •Chemotion-KG = ontology-driven, BFO-compliant chemistry knowledge graph •SPARQL CONSTRUCT + ODPs → modular semantic integration Ebrahim Norouzi et al., Chemotion Knowledge Graph, Sci-K 2025 @ ISWC, Nov 2nd, Nara, Japan
www.ditrare.de/en 21 ■Stay connected with DiTraRe ○Website: www.ditrare.de/en ○Email: [email protected] ○LinkedIn: www.linkedin.com/company/ditrare ○Mastodon: social.kit.edu/@DiTraRe ○YouTube: www.youtube.com/@DiTraRe ○Zenodo: zenodo.org/communities/ditrare ■Symposium on Digitalisation of Research: 2-3 December 2025, Karlsruhe ■Monthly Interdisciplinary Colloquium on Digitalisation of Research DiTraRe Leibniz Science Campus Digital Transformation of Research
ebrahim.norouzi@fiz-Karlsruhe.de https://sigmoid.social/@enorouzi https://ebrahimnorouzi.github.io/ Thank you very much for your attention! https://ditrare.ise.fiz-karlsruhe.de/chemotion-kg/