untitled

294 12 Building Bioinformatics Ontologies

anew. However, there are risks involved that must be balanced against the

advantages. Here are some of the reasons why existing ontologies might not

be appropriate:

1. The ontology may have inappropriate features or the wrong level of de-
   tail. This could happen because the ontology was constructed for a differ-
   ent purpose or in a different context.


2. The ontology may be in an incompatible language. For example, it is dif-
   ficult, in general, to convert a database schema to one of the XML ontol-
   ogy languages. Similarly, it is difficult to convert most XML DTDs and
   schemas to RDF or OWL because DTDs and schemas may not explicitly
   specify relationships. For example, the hospital that is developing the
   medical chart ontology may already have a relational database that in-
   cludes medical chart information. However, relational database schemas
   are difficult to convert to any of the XML ontology languages.


3. Existing ontologies, especially database schemas, can have artifacts that
   were introduced to improve performance rather than being fundamental
   to the domain. Aside from the fact that such artifacts are conceptually
   irrelevant, they might actually result in worse performance because per-
   formance is highly dependent on the environment in which the database
   is being used.

Having determined that an existing ontology is at least partially suitable

for reuse, there are a number of ways to incorporate it in another ontology.

The simplest technique is to download the ontology and copy (i.e., “cut and

paste”) all or part of it.^1 One can then modify it if necessary. However, this

technique can result in very large and unwieldy ontologies. Maintaining an

ontology developed in this way can become very time-consuming if several

ontologies have been reused.

There are two alternatives to copying:

1. Include.This is nearly the same as cutting and pasting except that it oc-
   curs every time that the document is processed. The inclusion is speci-
   fied by giving the URL of the ontology to be included. The ontology is
   downloaded and substituted like any other included document into the
   place where the inclusion was requested. An example of this is shown in
   section 1.4 where five XML documents containing experimental data are


2. Of course, one should be careful to ensure that doing this does not violate the copyright.