128
This part addresses how ontologies are constructed and used. One uses on-
tologies far more frequently than one creates them, and it is a good idea to
have some experience with how ontologies are used before attempting to
design new ontologies. Accordingly, this part begins with the many uses for
ontologies, and it ends with how one constructs them.
One of the most common uses of ontologies is for querying and retrieval.
The first three chapters discuss how query processing works and how to for-
mulate effective queries. Because ontologies have deductive capabilities, the
result of a query makes use of inferred information as well as explicitly spec-
ified information. There are two main points of view that one can take with
respect to retrieval. The first point of view is based on imprecise queries,
while the second point of view is based on precise, logical queries. Imprecise
bioinformatics queries can be expressed in two ways: natural language or
biological sequences. Chapter 6 considers natural language queries, while
chapter 7 deals with biological sequence queries. Chapter 8 introduces com-
puter languages for unambiguous queries.
After information retrieval, the most common activity involving ontolo-
gies is transformation. The process whereby information is transformed from
one format to another is surveyed in chapter 9. Such processes can have
many steps and involve many groups of individuals. It is helpful to under-
stand the entire transformation process so that the individual steps can serve
the overall process better.
The individual transformation steps use a variety of programming lan-
guages and tools. One of the most common is Perl. While Perl is especially
well suited for data transformations involving unstructured files, it can also
be used for structured data. Chapter 10 is an introduction to Perl that em-
phasizes its use for data transformations. While Perl can be used effectively
on XML documents, there is now a language specifically designed for trans-
forming XML. This language is called XSLT, and it is introduced in chap-
ter 11. As bioinformatics data migrate from flat files to XML structured files,
one can expect that XSLT will play an increasing role.
This part ends with a detailed treatment of the process whereby ontolo-
gies are built. The ontologies and databases that were surveyed in chapter 5
were substantial endeavors involving many individuals and requiring the
agreement of the community being served. While ontologies certainly can
be developed in this way, it is also possible for ontologies to serve smaller
communities for more limited purposes. Chapter 12 is a practical guide for
developing ontologies in a systematic manner, whether the ontology will be
used by a large community, a small community, or even a single individual.
