3.1: Deciding on reuse of existing ontologies
When should we consider reuse? Do actual candidate ontologies exist out there? How should we decide upon the most appropriate one?
Question: How to decide when to reuse and which ontologies to reuse?
Category: Conceptual
Problem owner: Linda
Date: 14 Dec 2020
Context
We do not want to reinvent the wheel when designing an ontology for a specific domain. So when feasible, we want to reuse already existing ontologies. But how do we decide when to reuse and which ontologies to reuse?
Various ways to reuse an existing ontology exist. It is important to decide on the best way to reuse an ontology based on the aspects above. More details on these various ways the can be found in the next chapter.
Problem
Three specific problems arise on deciding for ontology reuse:
Search and find relevant existing ontologies: you want to reuse ontologies, but: where do you search and find relevant ontologies?
Judge the quality of existing ontologies: which criteria should you consider when deciding whether or not to reuse an existing ontology?
Specification vs documentation: you might encounter differences between the actual specification of this ontology and its documentation: what you should take as the truth to build you own ontology?
Discussed options
Search and find relevant existing ontologies
Use of Google: With the right keywords, the first thing to do is just google for existing ontologies. However, keep in mind that this has its disadvantages, because you get an information overload, it is only useful when you know very specifically what you are looking for. Besides, some ontologies are not findable via Google.
Repositories of ontologies: Repositories do exist, although their consistency, their maintenance and the age of the ontologies they carry are all showing signs of the fact that we are dealing with new technology here. Despite these drawbacks, the mere fact of their existence and accessibility remains very valuable.
Generic
European Commission Core Vocabularies and EU Vocabularies
Indiana University, List of Ontologies
Ontology design patterns (for small reuse candidates)
Foundational ontologies
Wikipedia on Upper Ontologies, incl. a table of characteristics, originally taken from "A Comparison of Upper Ontologies" (Mascardi07)
Developed by TNO
Papers: you can search for papers that describe ontologies on this topic. Be sure to analyze the model/ontology to check whether it fits your topic precisely.
Well known and accepted ontologies:
To be completed
Judge the quality of the ontology
To judge the quality of an existing ontology and to decide whether or not to reuse the ontology, you might want to take these aspects into account:
Creation
Who was involved in the creation of the ontology?
Is the ontology the result of a purely academic exercise? Reusing the ontology can be useful, but assume no ontology maintenance.
Documentation
Is the ontology well-documented, i.e., are definitions of all concepts available? Are the definitions included in the ontological concepts?
Use
Is the ontology (re)used by others? If not, reconsider reusing this ontology or look for an alternative.
How generic is the ontology? More generic concepts might be better suitable for reuse than very specific concepts. But for specific applications, reusing specific ontologies can be useful.
What is the context of the ontology? Only reuse when this context fits the context of the purpose and application of your ontology.
Maintenance
Does maintenance actually take place? If not, consider the ontology for inspiration only as opposed to actual reuse. Or, when it is spot on applicable, consider to get in contact with the authors to negotiate who is going to maintain it; them or us.
If maintenance is in place, who maintains the ontology? What rules and procedures are in place that you require to know before reusing it, e.g., is version control assumed with versioning as part of the IRI's, what is the retention period for older versions?
Specification vs documentation
Some advices are:
It depends: depending on the type of difference you encounter, sometimes the specification and sometimes the documentation might be (more) correct.
Specification over documentation: take the model/specification as first truth, look at the documentation later.
Just ask: Send the authors of the ontology and e-mail. They actually answer most of the times.
Outcome
Search and find relevant existing ontologies
Are you working on an ontology in the medical domain? You're lucky, this domain is very well covered.
Are you working on modelling other domains? Google is your friend, with the right keywords. Also check the list of Ontology repositories (at Discussed options) and well known and accepted ontologies.
Judge the quality of the ontology
Study the creation, documentation, use and maintenance of the existing ontology, and compare the ourcomes with the purpose, business scenarios and/or CQ's for your ontology. In the end, these represent the primary criteria according to which the competencies of the (part of the) existing ontology should be judged. When it passes, additional criteria may be formulated about computational performance, its comprehensabilty, its accuracy and/or (un)necessary complexity, and particularly its maintainability since our ontologies are generally meant for operational use by customers, and hence evolution will apply.
References
When you want to read more about reuse of ontologies, these are good reads:
Lonsdale, D., Embley, D. W., Ding, Y., Xu, L., & Hepp, M. (2010). Reusing ontologies and language components for ontology generation. Data and Knowledge Engineering, 69(4), 318–330. https://doi.org/10.1016/j.datak.2009.08.003
Simperl, E. (2009). Reusing ontologies on the Semantic Web: A feasibility study. Data and Knowledge Engineering, 68(10), 905–925. https://doi.org/10.1016/j.datak.2009.02.002
Last updated