What is the difference between a Taxonomy and an Ontology?
From the post:
In the world of information management, two common terms that people use are “taxonomy” and “ontology” but people often wonder what the difference between the two terms are. In many of our webinars, this question comes up so I wanted to provide an answer on our blog.
When I first read this post, I thought it was an April Fool’s post. But check the date: March 15, 2013. Unless April Fool’s day came early this year.
After reading the post you will find that what the author calls a taxonomy is actually an ontology.
Don’t take my word for it, see the original post.
I think the difference between a taxonomy and an ontology is that an ontology costs more.
I don’t know of any other universal differences between the two.
I first saw this in Taxonomy or Ontology by April Holmes.
I usually find that people are quite confused about the definitions of the terms taxonomy, thesaurus, and ontology, and that post is no different.
Personally, I like the graphic at Marcia Zeng’s Controlled Vocabularies Primer:
http://marciazeng.slis.kent.edu/Z3919/3structure.htm
It uses the definition of “taxonomy” that I’m most familiar with; it limits them to only hierarchical relationships. This graphic does not include “ontology”, but IMO it would be to the right of “thesaurus”, and instead of “Associative relationships” — which is really only one kind of relationship, a generic “related term”, or RT for short — it would say something like “Vocabulary designer-defined relationships”. (Hmmm, maybe I should make my own version of this graphic.)
If we compare the WAND post with the graphic at Zeng’s primer, you can see that they’ve confused “taxonomy” with “thesaurus” and I think they do the reader a disservice by glossing over the difference between the infinite relationships in an ontology and the single type of related term relationship found in what they are calling a taxonomy. I do think the difference matters, because much of the software available for creating thesauri and controlled vocabularies will only give you the RT relationship to work with.
Lately my favorite post about ontology and taxonomy is this one:
http://www.vocabcontrol.com/?p=210
I particularly like it for this sentence: “One way of thinking about this is to see taxonomies as vertical navigation and ontologies as horizontal,” as well as the observation that ontologies can be used to link taxonomies together. I’d like to also add that when you link a set of disjoint taxonomies together, you effectively have a faceted classification system, which you can read more about here: http://eprints.soton.ac.uk/271488/
Comment by marijane — April 1, 2013 @ 11:08 pm
I was troubled by the “vertical vs. horizontal” distinction but your second link also says:
“Basically, an ontology is a knowledge model (like a taxonomy or a flow chart) that describes relationships between things. The main difference between ontologies and taxonomies is that taxonomies are restricted to broader and narrower relationships whereas ontologies can hold any kind of relationship you give them.”
I can live with that.
And you’re right, the difference is important. My glossing over the differences is from reading countless emails where ontology advocates tout their tools/services over other approaches because they have an “ontology.” Whatever its name, I prefer a tool suited to the task at hand.
Sometimes that may be an ontology in your sense of the word and at other times, a taxonomy may be all that is required.
PS: Thanks for the pointer to VocabControl! Looks really nice.
Comment by Patrick Durusau — April 2, 2013 @ 1:05 pm
I can see why that might be troubling, since an ontology arguably does both. And a thesaurus, too, to some extent.
I wonder if a graphic of the differences and similarities between the various controlled vocabulary types might be better expressed as a venn diagram?
Comment by marijane — April 2, 2013 @ 6:37 pm
Question: How would you represent the properties you assign to each type in a Venn diagram?
Thinking one approach to a discussion of controlled vocabulary types would be to list for each one the properties you take as identifying that type. Then other people can indicate which properties they agree/disagree with.
Perhaps reaching a consensus for a particular discussion and then representing that agreement with a Venn diagram?
Thinking starting with the Venn diagram leaves too much undiscussed/unspecified that could lead to disagreement.
Comment by Patrick Durusau — April 3, 2013 @ 2:06 pm
I wouldn’t. What I would want to show is how some vocabulary structures are wholly contained in others, ie, taxonomy and thesauri can be fully expressed in an ontology; individual facets can be represented as lists, etc. But it wouldn’t necessarily clear up disagreements about what the terms themselves mean, this is true.
Comment by marijane — April 3, 2013 @ 6:56 pm
How good are you with drawing tools?
Thinking a good diagram, even if it doesn’t solve all the issues, could be quite helpful.
As a matter of fact, I lean more towards good enough solutions these days than magnum opus ones. For one thing they get done quicker. 😉
Comment by Patrick Durusau — April 4, 2013 @ 3:22 pm
I am pretty good with Visio. I was thinking about this this morning and I can see a couple ways to potentially diagram this, both focusing on the properties of the various vocabulary structures and the differences between them as a whole… I will try to sit down and draw something in the next week or so.
Comment by marijane — April 5, 2013 @ 3:03 pm
Looking forward to the result!
Comment by Patrick Durusau — April 6, 2013 @ 5:26 am