Chemotion: Digital Chemistry Data Lifecycle

• Chemotion is a chemistry-specific digital management system combining an electronic lab notebook and a FAIR-oriented repository.
• It streamlines the entire experimental process from planning and documentation to curation and data reuse with structured metadata.
• Its semantic extension via a BFO-compliant knowledge graph enables AI-readiness, federated queries, and cross-domain reasoning.

Searching arXiv for the provided Chemotion-related papers to ground the article in current literature. Chemotion is a research data management environment developed at Karlsruhe Institute of Technology specifically for chemistry. It is described as comprising two tightly coupled components—Chemotion ELN, an open-source electronic lab notebook tailored to laboratory chemists, and the Chemotion Repository, a FAIR-oriented research data repository that interoperates with the ELN—and, in more recent work, as the basis of a BFO-compliant Chemotion Knowledge Graph that semantically lifts curated repository metadata into an ontology-driven RDF graph. In parallel, Chemotion is used pedagogically as a central instrument for teaching research data management and FAIR principles in an advanced inorganic laboratory course at RWTH Aachen University (Norouzi et al., 1 Sep 2025, Hoffmann et al., 23 Sep 2025).

1. System composition and operational scope

Chemotion is consistently presented as a chemistry-specific environment for the complete digital data lifecycle rather than as a standalone notebook. Chemotion ELN supports planning and documenting reactions and experiments, drawing and managing chemical structures via an embedded Ketcher editor, capturing conditions, reagents, samples, yields, and observations, handling analytical data including NMR, MS, IR, and Raman, and organizing experiments in projects with user management and local workflow support. The Chemotion Repository enables deposition of reaction information and analytical data from the ELN, curation and review processes for compound characterization data, publication of datasets with DOIs and rich metadata, and public access to reaction records, spectra, and associated metadata (Norouzi et al., 1 Sep 2025).

Component	Core functions	Role in the lifecycle
Chemotion ELN	Experiment planning and documentation; chemical structure management; metadata and analytical data capture	Plan, record, analyze
Chemotion Repository	Deposition, curation, review, DOI-based publication, public access	Curate, share, preserve, reuse

The lifecycle is explicitly described in process terms such as

$$\text{Plan} \rightarrow \text{Record} \rightarrow \text{Analyze} \rightarrow \text{Curate} \rightarrow \text{Share} \rightarrow \text{Reuse}$$

and, in the educational setting,

$$\text{Plan} \rightarrow \text{Capture (ELN)} \rightarrow \text{Curate} \rightarrow \text{Publish (Repository)} \rightarrow \text{Reuse}.$$

These formulations emphasize that Chemotion spans experiment planning, execution, documentation, curation, dissemination, and long-term preservation rather than merely replacing a paper notebook.

A recurring simplification is to describe Chemotion only as an ELN. The source material instead defines it as an ELN tightly linked to a repository, with the repository providing the publication-ready, curated metadata layer that later serves as the source for Chemotion-KG. This distinction is central for understanding both its FAIR orientation and its later semantic augmentation.

2. Data structures and laboratory workflow

In practical use, Chemotion organizes chemical work as structured digital objects. A typical workflow begins with planning and performing experiments in Chemotion ELN, including reaction setup, reagents, samples, chemical structures such as SMILES and InChI, and analytical measurements such as NMR or IR. Selected experiments are then exported or pushed to the Chemotion Repository, where metadata are curated and enriched, internal review is performed, and datasets are published with persistent identifiers. Other researchers can subsequently discover and reuse the material via repository search interfaces and APIs (Norouzi et al., 1 Sep 2025).

The teaching-oriented description makes explicit that the ELN is a web-based system in which chemists plan, run, and document experiments digitally instead of in paper notebooks. In the advanced inorganic laboratory course, students plan, document, and evaluate experiments digitally; document reactants, procedures, conditions, observations, and modifications in real time; and attach spectra, crystal structures, and other measurements directly to relevant sample or reaction entries. The chemical data model is described as supporting explicit specification of reaction schemes and sample entities, including identifiers for substances, stoichiometries, and relationships between reactions and samples (Hoffmann et al., 23 Sep 2025).

A compact formal description given for this organization is

$$\text{Experiment} \supset \{\text{Reactions}, \text{Samples}, \text{Analyses}, \text{Metadata}\}.$$

Each of these objects carries its own metadata fields, which may include chemical identifiers, provenance, instrument settings, and links to repository entries. The educational paper also stresses the inclusion of synthesis information, including negative results. This suggests that Chemotion’s data model is intended not only for successful syntheses or polished publications, but also for preserving procedurally and analytically complete records that remain reusable even when experimental outcomes are unfavorable.

The significance of this workflow lies in its replacement of inaccessible formats such as handwritten records or PDFs with structured, machine-readable data. The source material frames that shift as a prerequisite for long-term reuse, reproducibility, and machine-actionable downstream applications.

3. Semantic lifting into Chemotion-KG

The Chemotion Knowledge Graph is constructed from the repository side of Chemotion, specifically from curated, publication-ready metadata exposed through the Chemotion API in JSON-LD. The source schema relies primarily on schema:Dataset, schema:ChemicalSubstance, and schema:Study, covering (in)organic reactions, NMR, MS, IR, and Raman spectroscopy datasets, together with publication metadata such as creator, publisher, license, and URL. The pipeline periodically queries the API, retrieves JSON-LD descriptions, converts them into RDF while maintaining original schema.org semantics, and then transforms them into a BFO-aligned graph by SPARQL CONSTRUCT queries managed via the shmarql platform (Norouzi et al., 1 Sep 2025).

The motivation for this semantic lifting is the heterogeneity of chemistry data: complex workflows, multi-level entities, complex provenance, and numerous data formats and devices. The knowledge-graph work explicitly states that schema.org is suitable for lightweight web interoperability but limited for deep provenance modeling, expressing research lifecycle stages, capturing rich structural semantics for experiments and substances, and aligning with upper-level ontologies required for cross-domain AI reasoning. The resulting semantic layer is therefore intended to move from “metadata for web pages” to a BFO-aligned ontology-driven knowledge graph.

The transformation step recasts schema.org classes into NFDICore, BFO, and ChEBI classes. Representative mappings include schema:Dataset to nfdicore:NFDI_0000009, schema:Person to nfdicore:NFDI_0000004, schema:Organization to nfdicore:NFDI_0000003, schema:Study to obo:BFO_0000015, and schema:ChemicalSubstance to obo:CHEBI_59999, plus obo:CHEBI_23367 where appropriate. Properties such as schema:creator, schema:publisher, and schema:measurementTechnique are likewise aligned to NFDICore relations. Titles, descriptions, URLs, and identifiers are remodeled as separate information content entities rather than bare literals, with newly minted IRIs under https://ditrare.ise.fiz-karlsruhe.de/chemotion-kg/nodes/….

The materialized RDF is stored in named graphs organized by submission date, and IRIs encode date and Chemotion identifiers. Data ingestion is described as daily. The knowledge graph is hosted by FIZ Karlsruhe – Information Service Engineering, with a public base at https://ditrare.ise.fiz-karlsruhe.de/chemotion-kg/, a SPARQL endpoint at https://ditrare.ise.fiz-karlsruhe.de/chemotion-kg/sparql, and source code at https://github.com/ISE-FIZKarlsruhe/chemotion-kg.

As of July 2025, Chemotion-KG is reported to contain 1,462,187 RDF triples and 87,782 instantiated entities, including 20,701 datasets, 20,563 studies, 3,746 molecular entities, 4,923 generic chemical substances, and 250 individual creators with explicit provenance. These figures do not constitute an evaluation of reasoning quality or extraction accuracy; the source explicitly notes that precision, recall, and reasoning-performance metrics are not presented. Their evidentiary role is instead to demonstrate operational scale and content diversity.

4. Ontological design patterns and formal representation

The ontological foundation of Chemotion-KG is Basic Formal Ontology. The knowledge-graph work distinguishes continuants—agents, chemical substances, molecular entities, roles, and information content entities—from occurrents or processes such as studies, publishing events, and laboratory workflows. Representative mappings are stated explicitly: $\text{Study} \sqsubseteq \text{BFO\_0000015},$

$\text{Dataset} \sqsubseteq \text{IAO\_0000100},$

$$\text{Person} \sqsubseteq \text{Agent} \sqsubseteq \text{BFO independent continuant}.$$

Chemical entities are aligned to ChEBI through

$$\text{ChemicalSubstance} \sqsubseteq \text{CHEBI\_59999}$$

and

$$\text{MolecularEntity} \sqsubseteq \text{CHEBI\_23367}.$$

A central ontology design pattern in the graph is the Process–Agent–Role pattern. Let $P$ be a process, $A$ an agent, and $$\text{Plan} \rightarrow \text{Capture (ELN)} \rightarrow \text{Curate} \rightarrow \text{Publish (Repository)} \rightarrow \text{Reuse}.$$0 a role. The relations used are bfo:has_participant, bfo:bearer_of, and bfo:realizes, linking participation in a process to a role borne by an agent. The source gives an OWL-style axiom pattern: $$\text{Plan} \rightarrow \text{Capture (ELN)} \rightarrow \text{Curate} \rightarrow \text{Publish (Repository)} \rightarrow \text{Reuse}.$$1 and

$$\text{Plan} \rightarrow \text{Capture (ELN)} \rightarrow \text{Curate} \rightarrow \text{Publish (Repository)} \rightarrow \text{Reuse}.$$2

This pattern is used to represent persons as agents bearing roles such as author, corresponding author, or principal investigator, with studies or publishing processes realizing those roles (Norouzi et al., 1 Sep 2025).

Studies are modeled as processes and linked to datasets, substances, and publishing processes. Publishing is represented by nfdicore:NFDI_0000014 and linked both to temporal regions and to standards. This introduces explicit lifecycle structure into the graph. In addition, qualities and measurements are modeled in BFO/IAO terms. Molecular weight, for example, is described as a quality with associated measurement datum and unit: $$\text{Plan} \rightarrow \text{Capture (ELN)} \rightarrow \text{Curate} \rightarrow \text{Publish (Repository)} \rightarrow \text{Reuse}.$$3 with relations

$$\text{Plan} \rightarrow \text{Capture (ELN)} \rightarrow \text{Curate} \rightarrow \text{Publish (Repository)} \rightarrow \text{Reuse}.$$4

The conceptual consequence is a formally typed representation of chemical research in which datasets, studies, agents, roles, chemical entities, temporal regions, and standards are not merely adjacent metadata fields but explicit nodes and relations. A plausible implication is that this design supports cross-domain reasoning in a way that schema.org-only descriptions do not.

5. FAIR semantics, querying, and AI-readiness

Chemotion is explicitly aligned with the FAIR principles, and Chemotion-KG extends this with what the source calls semantic FAIRness. Findability is supported by persistent URIs that encode submission date and Chemotion identifiers, stable URIs for information content entity nodes, and indexing of datasets, studies, chemical substances, and creators via ontology-based types. Accessibility is provided through public hosting by FIZ Karlsruhe – Information Service Engineering, a public SPARQL endpoint, and openly available source code and pipeline scripts. Interoperability is grounded in RDF together with BFO, NFDICore, ChEBI, and preserved schema.org semantics. Reusability is supported by documented ontology design patterns, clear class and property hierarchies, provenance and lifecycle relations, and planned SHACL validation (Norouzi et al., 1 Sep 2025).

The educational account presents the same FAIR orientation from the operational side. Chemotion makes experiments, reactions, samples, and datasets findable through structured metadata and clear identifiers; accessible through long-term digital storage governed by licenses and policies; interoperable through machine-readable formats compatible with chemistry infrastructure such as NFDI4Chem tools and knowledge graphs; and reusable because experiments are fully described and instrument data is linked and annotated. The source offers the conceptual expression

$$\text{Plan} \rightarrow \text{Capture (ELN)} \rightarrow \text{Curate} \rightarrow \text{Publish (Repository)} \rightarrow \text{Reuse}.$$5

without proposing a formal metric (Hoffmann et al., 23 Sep 2025).

The knowledge-graph work frames Chemotion-KG as an AI-ready substrate. Because agents, roles, processes, datasets, and substances are formally typed and linked, reasoners can infer who participated in which experiment and under which role, which datasets are outputs of which studies, and how publishing processes relate to standards and time intervals. The integrated graph of structural data such as InChI and SMILES, analytical spectra, and experimental metadata is described as enabling reaction similarity search, automatic classification of experiments by type or technique using nfdicore:NFDI_0000216, and identification of workflows and typical sequences of processes.

The same work also connects Chemotion-KG to future integration with ML, LLMs, and autonomous experiment planning. The stated direction is to use Chemotion-KG as a structured knowledge base to train or condition ML models, to integrate LLMs with structured Chemotion-KG data for advanced question answering and knowledge extraction, and to support autonomous discovery and self-driving laboratories through machine-interpretable data and workflows. The source does not present complete AI demonstrations; its claim is infrastructural rather than benchmark-driven.

6. Pedagogical use and the DALIA training ecosystem

At RWTH Aachen University, Chemotion is integrated into a fifth-semester advanced inorganic chemistry laboratory course as the central environment for experimental documentation and the practical teaching of research data management. Students are expected to use Chemotion for all experimental documentation, and the course explicitly includes FAIR, TRUST, O3, metadata, and data lifecycle concepts among its learning objectives. In this setting, Chemotion functions as the operational environment in which abstract RDM principles become concrete through experiment planning, metadata capture, repository upload, and visualization of the full data life cycle (Hoffmann et al., 23 Sep 2025).

The pedagogical design is multi-component. A seminar introduces the importance of RDM, FAIR principles, and the role of ELNs and repositories in modern chemistry. Teaching videos based on Chemotion’s official documentation cover basic ELN functions and course-specific workflows. Interactive Moodle elements reinforce metadata quality, legal aspects, and data reuse, and a final RDM test assesses understanding. Assessment is both formative, through interactive modules and surveys, and summative, through the final test and evaluation of the quality of Chemotion entries, including completeness of data, correctness of metadata, and ability to reproduce experiments.

DALIA plays a complementary role as a discovery platform for the instructional materials associated with Chemotion rather than for experimental data themselves. It is described as a dedicated OER platform that indexes educational and training resources using a metadata-rich model, the DALIA Interchange Format, implemented with RDF and Schema.org educational resource elements. The formal description given is

$$\text{Plan} \rightarrow \text{Capture (ELN)} \rightarrow \text{Curate} \rightarrow \text{Publish (Repository)} \rightarrow \text{Reuse}.$$6

In the case study, DALIA indexes the Chemotion ELN instruction videos, allowing discovery through faceted search by discipline, target group, and resource type.

This educational deployment clarifies another potential misconception: FAIR implementation in Chemotion is not confined to repository publication after research is complete. In the course design, FAIR-oriented practice begins at experiment planning and data capture. The ELN is therefore both a data-management system and a didactic device for inculcating machine-readable, metadata-rich laboratory habits at the bachelor level.

7. Infrastructure context, limitations, and prospective development

Chemotion is linked to broader research-data infrastructures. It is described as an open-source, DFG-funded tool tightly linked to the German National Research Data Infrastructure for Chemistry, NFDI4Chem. The educational work places it alongside minimum information standards for chemistry data, FAIR and TRUST principles for repositories, and the O3 guidelines for sustainable curated resources. The knowledge-graph work situates Chemotion-KG within the Leibniz Science Campus “Digital Transformation of Research” (DiTraRe), specifically the AI4DiTraRe dimension “Exploration and Knowledge Organisation,” where Chemotion serves as a use case for studying the impact of digitalization and AI on research workflows and data cultures (Norouzi et al., 1 Sep 2025, Hoffmann et al., 23 Sep 2025).

Within the chemistry knowledge-graph landscape, Chemotion-KG is distinguished in the source material by its use of upper-level ontology BFO, ontology design patterns for Process–Agent–Role and for datasets, studies, and chemical entities, and the combined use of NFDICore, ChEBI, and schema.org within a coherent semantic framework. Planned integration targets include PubChem RDF, ChemSpider, and the NFDI4Chem Knowledge Graph. The expected result is support for federated queries and cross-resource reasoning.

The current system also has explicitly stated limitations. The present Chemotion-KG focuses on repository-level metadata and does not yet cover the full Chemotion model. Future work is intended to incorporate more detailed dataset-level information, richer descriptions of chemical substances, including roles in reactions, conditions, and stoichiometry, and SHACL shapes for validation of RDF graph structure. External linking remains limited so far. The knowledge-graph paper further points to additional automation of curation workflows in Chemotion and expanded integration with ML and LLMs for semantic alignment and advanced query answering (Norouzi et al., 1 Sep 2025).

These limitations are not framed as contradictions of the existing system but as boundary conditions of the present implementation. The combined record across the two cited works presents Chemotion as an operational chemistry data environment whose ELN and repository already realize a complete digital data lifecycle, and whose ongoing semantic extension into Chemotion-KG aims to make those data not only FAIR in a repository sense but also formally interoperable, provenance-rich, and directly usable in ontology-based reasoning and AI-supported chemistry.