The Helmholtz Association is actively digitizing research outcomes to drive progress and innovation. The vast volumes of digital data are diverse in terms of their formats and the semantic descriptions used in their interchange, and storage. Therefore, a semantic frame of reference is required to facilitate interoperability throughout the Helmholtz digital ecosystem. This paper presents the Helmholtz Digitization Ontology (HDO), which is intended to serve that purpose. HDO is a mid-level ontology that contains concepts representing digital assets relevant to the Helmholtz digital ecosystem, data creation, management, and exchange. It is developed within the framework of the Helmholtz Metadata Collaboration (HMC) with contributors from various scientific backgrounds. HDO serves as a harmonized semantic framework and machine-actionable reference across all Helmholtz research fields.
The Helmholtz Association comprises 18 research centers operating across Germany,
focusing on a wide range of scientific topics and methods within six research areas fields 1. The
Helmholtz Metadata Collaboration (HMC)2 is an association-wide operating platform that
supports (meta)data harmonization and information engineering across all research fields intending
to make data within Helmholtz adhere to the FAIR principles [
Motivation and requirements. The heterogeneity of scientific contexts within Helmholtz
leads to ambiguity and conflicts regarding metadata semantics, e.g. in developed metadata
schemas, tools or general communication between collaborators, when data is exchanged in
interdisciplinary collaboration. Due to their ability to establish clear context and relationships
between concepts [
Objectives. The main objectives of the Helmholtz Digitization Ontology (HDO) are: 1)
Creating a standardized semantic framework with terminology that can reduce semantic uncertainty
and ambiguity and thereby increase semantic interoperability between various Helmholtz
systems. 2) Facilitating data integration in diferent Helmholtz systems, e.g., the institutional
Helmholtz Knowledge Graph [
In HDO, we provide well-defined concepts with rigorous semantic and unambiguous definitions. Class definitions aim to: 1) outline the intrinsic characteristics of the term being defined, 2) avoid circularity, 3) be neither excessively broad (to avoid ambiguity), nor too narrow (to allow implementation of further sub-classes where necessary), and 4) be easily understood using common, unambiguous terminology, which is important, especially concerning HDO’s purpose of serving as a mid-level ontology that provides common understanding and reduces miscommunication across diferent research fields.
To create class definitions, we follow Aristotelian logic, specifically, definitions adhere to the genus-diferentia form 3. Genus-diferentia definitions follow the form: “A (the class label) is a B (the genus or superclass) which is C (the differentia)”. For instance, consider the definition of “ JSON file”, which is “A file which conforms to JSON format”. Here, the genus part is “file”, which is the superclass of JSON file, from which all diferentia and properties are inherited. The remaining part is the diferentia, which comprises the features distinguishing the currently defined term from its genus and siblings.
The key aspects of HDO development are illustrated in Figure 1. The development is carried out in three phases: initialization, implementation, and adoption and adaption.
1. Initialization phase: In the early stages of the development, an internal GitLab repository was created to gather a set of core terms and their definitions in per-term YAML files. We followed a template with keys such as definition, synonyms, comments, and seeAlso. 3https://en.wikiversity.org/wiki/Dominant_group/Genus_diferentia_definition creator date modified date created contributor Class Axioms LODE license editor note
BFO RO DC FOAF definition
label comment
Metadata
Logic IAO Reused ontologies
Class annotations
Documentation
Development
ODK OWL
PID full provenance acronym
Bilingual broad synonym related synonym
English
German cross reference examples gloss plural exact synonym
2. Implementation phase: After collecting terms, YAML files were converted and
merged into one OWL file, where keys of the template were mapped onto existing and
imported annotation properties (e.g. definition was mapped to iao:definition).
Further development was carried out in the OWL file in a public repository 4, and managed
using the Ontology Development Kit [
3. Adoption and Adaption: Upon publication of HDO, it will be used in use cases across the diferent Helmholtz research fields. This will test the ontology against use case-specific requirements and allow further adaption based on iterative exchange.
We focused on reusing classes from existing, well-known semantic artifacts, wherever possible, to ensure semantic interoperability. HDO is top-level aligned with the Basic Formal Ontology (BFO)6 to ensure interoperability with other mid- and domain-level ontologies. Furthermore, classes and properties from well-known Open Biological and Biomedical Ontologies (OBO)
entity occurrent continuant
generically dependent continuant specifically dependent continuant information structured data data were re-used, including: Information Artifact Ontology (IAO)7 (iao:action specification, iao:information content entity, iao:plan specification) and the Relation Ontology (RO)8 (ro:input of, ro:has characteristic and ro:has input).
An overview of the HDO core classes is shown in Figure 2. HDO establishes semantics for the core concepts of digital infrastructure, digital information management, and processes in data management and exchange. For example, HDO ofers a rigorous semantic context for the aspects of the FAIR principles. These concepts are modelled as bfo:disposition that inhere in hdo:data (HDO_00000009) either according to a practical understanding of hdo:findability, as well as specified according to the FAIR principles as hdo:findability according to the FAIR principles. Further, we created classes for diferent aspects of metadata standardization under hdo:information and aligned this with IAO classes (e.g., iao:information content entity). 3.3. Logic We asserted pairwise disjointness between mutually disjoint classes, e.g., hdo:digital infrastructure is disjoint with hdo:hardware. We used bfo:role and bfo:realizes to create a pattern that allows populating certain classes by inference. For example, the class hdo:agent is “An entity which realises an agent role”. This is inferred through hdo:agent
role and bfo:realizes. A similar example is the class hdo:tool which is defined as “A continuant which realises a tool role”. For this, as well as to allow reasoning about relevant concepts, several OWL axioms have been asserted in HDO. The EquivalentClasses axiom allows to state that several class expressions are equivalent to each other. For instance, the class hdo:data has the following axiom assertion: ’generically dependent continuant’ and
(’output of’ some (’encoding process’ and (’has input’ some signifier))) The SubClassOf axiom allows to state that each instance that fits a class expression is also an instance of that class. For instance, the class hdo:structured vocabulary has the following SubClassOf axiom assertion: vocabulary and ’has quality’ some structuredness
The Helmholtz Digitization Ontology was developed with the objective of providing harmonized
semantics as a reference framework for the Helmholtz digital ecosystem. The development
of HDO was open and transparent, and its full provenance is recorded. Further, we follow an
established ontology development framework (e.g., ODK) and align HDO with well-established
semantic frameworks in order to increase acceptance towards domain-level re-use and
application. One of the further use cases will be the semantic representation of FAIR digital
objects (FDOs) that will allow data integration between FDOs and the HMC Helmholtz KG
[