Definitions in Biomedical Ontologies: The Method ONTODEF

Textual definitions in ontologies are essential for the consistent use of specialized terms in interdisciplinary communities. Well-formulated definitions ensure consistency of human reasoning and support computational reasoning. These are requirements for an ontology to participate in biomedical repositories, such as the OBO Foundry [1-4].

In this article, the application of the OntoDef method for formulating textual definitions in biomedical ontologies is introduced and demonstrated. The method consists of systematizing principles to create definitions in natural language. It is developed within Information Science research and based on Aristotelian principles to define [5,6].

For Aristotle, each species of a “Genus” has a unique “Differentia” [7]. For example, the species “man” comprises the genus “animal” and a Differentia concerning other animals. Genus and Differentia define a term by providing the essence of an entity [8]. An Aristotelian definition is expressed in the form “α =def. β”, where α is the specie and β is the combination of the Genus + Differentia correspondent to α [7,9].

Principle of the unique inheritance

In ontologies, definitions are arranged hierarchically so that a term must not have two parents, that is, entities of a direct superior hierarchy. In having two parents we will incur multiple inheritance, which is not suitable for ontological definitions [5].

Principle of necessary and sufficient conditions

If a’ is a necessary condition for an entity to be of type A, then every entity that is of type A will have the condition a’ [8].

Principle of non-circularity

One should not define the term recursively; for example, laboratory examination is one performed in a clinical laboratory [10].

Principle of Intelligibility

The definition must use intelligible terms that are easy to understand so that the definition is not restricted to domain experts [5,10].

The OntoDef method is depicted in Figure 1. A full example of the Ontodef method to define a real term is presented in Table 1.

Figure 1: Algorithm for applying the OntoDef method. Source: Adapted from Souza AD [5] Souza AD et al. [6].

Table 1: Formulation of definition for the term “Acute erythroid leukemia” in Blood Ontology using OntoDef

The application of the OntoDef method brought a relevant contribution by addressing gaps in the way of creating definitions in ontologies. The method provides a view of the entity’s essence by establishing the necessary and sufficient conditions through methodological steps that condition the ontology creator to a reflection on the definition process.

1. IWOOD (2014) International workshop on definitions in ontologies.
2. Arp R, Smith B, Spear AD (2015) Building ontologies with basic formal ontology, MIT Press, Cambridge, USA.
3. Schlegel D, Seppälä S, Elkin P (2016) Definition coverage in the OBO foundry ontologies: The big picture.
4. Stevens R, Malone J, Williams S, Power R, Third A (2011) Automating generation of textual class definitions from OWL to English. Journal of Biomedical Semantics 2(Suppl 2): 1-20.
5. Souza AD (2015) Systematizing of the methodology of creating formal definitions in biomedical ontologies: An investigation in acute myeloid leukemia domain. Dissertation (Master of Information Science) School of Information Science, Federal University of Minas Gerais, Brazil.
6. Souza AD, Almeida MB (2016) Natural language definitions for the leukemia knowledge domain. Proceedings of the International Conference of Biomedical Ontologies, Corvallis, USA.
7. Berg J (1983) Aristotle's theory of definition, pp. 19-30.
8. Seppälä S, Schreiber Y, Ruttenberg A (2014) Textual and logical definitions in ontologies.
9. Seppälä S, Ruttenberg A, Schreiber Y, Smith B (2016) Definitions in ontologies. Notebooks of Lexicology 2(109): 173-205.
10. Smith B (2013) Introduction to the logic of definitions. International Workshop On Definitions In Ontologies, pp. 1-2.