Ontologies

1. Consider reusing existing ontologies to build upon established knowledge structures. #

IDENTIFY PLAN

Numerous ontologies have been developed for cultural heritage and humanities disciplines in recent years. Before creating your own ontology from scratch, examine existing ones to evaluate whether and how they could serve your research needs. This approach not only speeds up development but also results in a more robust final product.

Domain-relevant models, ontologies and vocabularies, used on a national and international level: CIDOC-CRM, FOAF, Schema.org, Dublin Core, LRMOO, DataCite, DCAT, SPAR Ontologies, and ArCO.
To explore existing models, use the LOV (Linked Open Vocabularies) database, BARTOC (Basic Registry of Thesauri, Ontologies and Classifications) or the H-SeTIS (Heritage – Semantic Tools and Interoperability Survey) database specific for the heritage domain.

2. Engage the user community and domain experts throughout the design phase to create more robust and relevant ontologies. #

PLAN PRODUCE

The first step in creating an ontology is defining its domain—the scope of reality it must represent—and identifying “competency questions.” These are user-oriented questions that help scope the ontology by determining what users want to learn when exploring and querying the ontology and its knowledge base.

Creating an effective ontology requires deep domain knowledge. Involving domain experts and community members early in the process helps identify precise terminology and appropriate properties to describe domain entities, their relationships, and competency questions. This collaborative approach ensures the ontology will be both useful and reusable.

Surveys and focus groups serve as effective tools for gathering domain expert feedback.

3. Adhere to shared methodologies for ontology design and implementation to ensure consistency and best practices. #

PLAN PRODUCE

The literature offers several established methodologies and best practices for creating ontologies:

Ontology Development 101: A guide to creating your first ontology, a comprehensive guide for beginners learning to model domains through ontologies;
Simplified Agile Methodology for Ontology Development (SAMOD), an advanced approach developed by the SPAR Ontologies team;
LOT (Linked Open Terms) Methodology an industrial method for developing ontologies and vocabularies, that encompasses four main phases: requirements specification, implementation, publication and maintenance;
The NeOn Methodology for Ontology Engineering proposes nine scenarios for building ontologies and networks, focusing on resource reuse, reengineering, and collaborative development.

When facing common challenges during domain modeling, you can apply Ontology Design Patterns as proven solutions.

For the practical development of ontologies, several graphical editors are available, including Protégé, Fluent Editor, and OWLGrEd. Antoher popular development platform for managing OWL ontologies is VocBench. For more options, see BARTOC's comprehensive list of editors and development environments.

Standard formats

As the primary developer of Semantic Web technologies, W3C has created several standard formats for expressing ontologies:

RDF XML (Resource Description Framework XML Syntax) serves as the foundation for many ontologies in the Semantic Web. It uses XML syntax to express relationships through triples (subject, predicate, object).
OWL (Web Ontology Language) represents complex knowledge about things and their relationships. Based on computational logic, it enables programs to verify consistency and reveal implicit knowledge.
Turtle provides a simplified, human-readable syntax for RDF that is more concise than RDF-XML.
JSON-LD (A JSON-based Serialisation for Linked Data) integrates Linked Data into web environments through JSON compatibility.

Other important W3C Semantic Web Standards include:

SKOS (Simple Knowledge Organization System) for creating vocabularies and taxonomies;
SPARQL (SPARQL Query Language for RDF) for querying diverse RDF data sources and retrieving both result sets and RDF graphs.

Guidelines for producing FAIR ontologies

The scientific literature offers several publications with clear, precise guidelines for producing FAIR ontologies. These cover essential topics like prefix and namespace conventions, as well as documentation publishing methods, including:

Garijo, Daniel, and María Poveda-Villalón. Best Practices for Implementing FAIR Vocabularies and Ontologies on the Web. arXiv, 2020, https://doi.org/10.48550/ARXIV.2003.13084.
Hugo, Wim, et al. D2.5 FAIR Semantics Recommendations Second Iteration. Dec. 2020, https://doi.org/10.5281/ZENODO.4314320.
Poveda-Villalón, María, et al. “Coming to Terms with FAIR Ontologies.” Knowledge Engineering and Knowledge Management, edited by C. Maria Keet and Michel Dumontier, vol. 12387, Springer International Publishing, 2020, pp. 255–70, https://doi.org/10.1007/978-3-030-61244-3_18.

To evaluate whether your ontology adheres to the good practices outlined in these publications, you can use tools such as O’FAIRe (Ontology FAIRness Evaluator) and FOOPS! (Ontology Pitfall Scanner for FAIR).

Another effective tool designed specifically for SKOS vocabularies is SKOS Play!, which enables users to validate and convert SKOS files to PDF and HTML formats, and also generates SKOS files from Excel spreadsheets.

4. Foster interoperability and facilitate integration by producing a comprehensive mapping that aligns your ontology with existing models and ontologies. #

PRODUCE

Conceptual alignment involves associating each concept expressed in your ontology — whether as a class or a property — with an equivalent concept in other models. This process enables the production of a comprehensive mapping, enabling the easy integration of data modelled according to your ontology into other datasets. Such a mapping is also useful for facilitating data maintenance and handling legacy data.

Many W3C models, such as DCAT, include mapping recommendations.
Standards and references:
Article: “Moving towards FAIR mappings and crosswalks”.

5. Provide users with detailed documentation, complete with practical usage examples and intuitive graphical representations of the ontology. #

PRODUCE DISSEMINATE

Documentation should be complete and accurate, beginning with clear definitions of all ontology elements. These definitions can incorporate references to external sources and specific examples to clarify concepts and properties.

To clearly demonstrate the ontology’s utility, include practical examples showing how it represents real-world objects familiar to your community.

Web environment for visualising and customising ontology documentation: LODE, WebVOWL, Widoco. List of visualisation tools.

Graphical representations effectively showcase your ontology’s classes and their relationships. You can produce simple UML class diagrams or use specific tools for ontologies such as Graffoo and Fluent Editor.

6. Ensure the permanence of URIs, considering the use of services for long-term stability. Conduct regular maintenance checks to ensure the continued functionality of links to your ontology. #

DEPOSIT

Ensuring your ontology’s long-term accessibility and reusability depends critically on functioning URIs. Regular maintenance checks are essential to keep the ontology ready to use

W3id, developed by the W3C Permanent Identifier Community Group, provides secure, permanent URL redirection services for web applications. As a result, the ontology maintains its stability and reliability, making it more appealing for reuse.

7. Store the ontology in multiple formats within certified repositories to guarantee easier access and long-term preservation. #