A More Elaborate Example


Let's delve into the realm of mushroom harvesting to make this more tangible (or fungible, if you can pardon a bad pun).

Al's Organic Mushrooms

Al harvests and farms wild and domestic organic mushrooms. He has a Web site and sells his mushrooms. Al is a dedicated mycophagist (mushroom eater), and he is committed to the mycologic ontology as shown in Figure 14.8. (Mycology is the branch of botany that studies fungi.) The mycologic ontology has a wealth of knowledge about mushrooms, but Al uses it only to get the names right. So Al can refer to his mushrooms as porcino, with the knowledge that he is synched up with the mycologic ontology entry for Boletus edulis.

click to expand
Figure 14.8: Al's ontology commitment.

Al also is committed to the California Produce Growers ontology, which helps him categorize his produce as fresh or dried and organic or nonorganic. It also specifies what restrictions there might be on transporting it.

Genus, Species, Family, Order, and Phylum

The mycology ontology has a wealth of data on mushrooms, but only for the fungi kingdom. The mycology ontology in turn commits to a more general biologic ontology that ensures it is using the standard nomenclature for the scientific naming of the many species.

The mycology ontology has information about whether each mushroom is poisonous or could cause hallucinations. It has a wealth of data to help identify what species a given mushroom is. Some of the tests are quite specific.

Conceivably someone could query Al's offering, when joined with the mycology page, and determine if any of the produce was poisonous, even if Al didn't know. If another mycology site committed to the same biologic ontology and knew (or suspected) that a particular species was poisonous, it could reveal that, even if the mycology ontology didn't know.

Food-Drug Interactions

It doesn't have to stop there. We did some work with Multum,[120] a company that has a comprehensive drug reaction database. Not only does it list drug–drug contraindications (which drugs should not be taken with which other drugs), but also drug–food and drug or food and biologic function (i.e., whether a drug or food will have an adverse reaction in some people based on their body chemistry).

Multum is going to commit to HCPCS (Figure 14.9) or an equivalent standards organization for identifying drugs, as well as to unambiguously identify the food substances that people might be allergic to. They commit to the biologic ontology and we're on our way. Now we can traverse the ontologies and determine that the drug warfarin is contraindicated for Boletus edulis. (I'm just making this up, so you warfarin takers need not panic, just yet.)

click to expand
Figure 14.9: A hospital and a drug interaction database.

This is just data, until we link your medical record, kept in this example at St. Jude's Hospital, to both the drug ontology (which all hospitals commit to) and the Multum database. Now we can connect your medical record, the drug warfarin, and Boletus edulis. Connect to Al's site, and it may have saved your life.

I've made it sound easier than it really is, currently (by about an order of magnitude or two). But this is conceivably possible, and this is where this technology is headed. The only drawback is that the process might be slow, which is why you are far more likely to have an agent do this research than a query.

My motivation in going through this example is to show how each local participant, committing only to a few ontologies, could be completely interlinked, and information could be deduced without anyone in the chain having the whole picture.

[120]See http://www.multum.com/ for further information.




Semantics in Business Systems(c) The Savvy Manager's Guide
Semantics in Business Systems: The Savvy Managers Guide (The Savvy Managers Guides)
ISBN: 1558609172
EAN: 2147483647
Year: 2005
Pages: 184
Authors: Dave McComb

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net