292 12 Building Bioinformatics Ontologies
intended to be used by a larger community than just a local group of people.
If information is going to be posted on the web, then it is especially important
to have a formal ontology of some kind.
Unfortunately, even if one is an expert in a field it can still be an effort to
formalize one’s knowledge about the subject. The two main ways to acquire
domain knowledge are glossaries and examples of use. For ontology devel-
opment projects, the statement of purpose or project requirements can also
be a source of relevant domain knowledge. Aglossaryordictionaryis a listing
of the terminology in the domain, together with definitions. Ausage exam-
pleis a test case for an ontology. It is a statement using terminology from
the field. It is best to express usage examples using simple statements in the
form of subject-verb-object. More complex statements that have subordinate
clauses are also useful, provided that the subordinate clauses are in the form
of subject-verb-object. Glossaries and usage examples are often combined. A
definition is an example of a use of the term, and it is common for a glossary
to include examples of how each term is used.
To understand the structure of the definition of a word, recall that in biol-
ogy the genus is the next most general concept above the species level. One
“defines” a species by giving its genus and then specifying the features of
the species that distinguish it within its genus. Words are defined the same
way. One first gives the next most general term, which is called the “genus
proximus,” and then one specifies the features of the word that distinguish it
within the genus proximus. Consider this definition of “lumbar puncture:”A lumbar puncture is a procedure whereby spinal fluid
is removed from the spinal canal for the purpose of
diagnostic testing.The genus proximus in this case is “procedure.” The subordinate clauses
“spinal fluid is removed from the spinal column” and “for the purpose of
diagnostic testing” distinguish “lumbar puncture” from other “procedures.”
While the ideal form for a statement is subject-verb-object, it is not al-
ways easy or natural to express domain knowledge this way. One sometimes
uses verbs as concepts, and nouns or noun phrases are sometimes necessary
for expressing a relationship. In the lumbar puncture definition one finds
the phrase “for the purpose of diagnostic testing.” To express this using
the subject-verb-object format, one must use a statement such as “Lumbar
puncture has the purpose of diagnostic testing.” Although this statement is
awkward, it is much better for ontology development.