Ontologies & Taxonomies

Ontologies & Taxonomies

At the KM World event recently I was asked (as was my co-presenter Dr Rao, and moderator Hugh Mckellar) what was the difference between an Ontology and a Taxonomy.

It was the kind of question you always hope somebody does not ask at a conference as it is difficult to answer. Though all three of us struggled to come up with a proper and clear definition on the fly, our combined efforts got it sort of right, and hopefully at least did not mislead the questioner further.

But it was in fairness a good question and one worthy of a proper answer as there is a huge amount of confusion around both of these concepts. So for purposes of clarity:

• Ontology: A model of the semantic and conceptual inter-relationships between various data, to encourge and support exchange, re-use and reduce duplication. Ontologies are built by computer scientists or technical architects and are as much focused on the connections between various software and related data as on the content of the data itself.

• Taxonomy: Are models focused on the document level, providing navigable hierarchies and common understandings of terminology for filing purposes. A good taxonomy provides the means to automatically understand where new content should be filed, and where to find older content all of this often automated.

(There are a range of software tools on the market aiming to provide automation of the taxonomy and ontology building exercise.)

In many regards they are quite different constructs, but they feed one another directly, or at least they should. For the reality is that even in the best funded enterprises one seldom finds a mature taxonomy in place, let alone an ontology.

For information management taxonomies are the bedrock of the information infrastructure, without them in place you have a map but no compass. You can find things, but only with difficulty and it is incredibly easy to lose things. Taxonomies provide a logical, and user friendly structure or tool or find and retrieve information easily and quickly, and to ensure newly acquired or created information is 'filed' properly.

Though good taxonomies make a lot of sense, they require discipline and they also generally require cross departmental support and processes to ensure they work and continue to work.

If you are a mid to large enterprise looking at a DM/WCM/KM or ECM project now or in the near future, starting off by building a solid taxonomy from the get-go will set you way ahead of the curve.

Ontologies on the other hand are quite frankly the latest fad - its not that they don't or can't work, or that one shouldn't build them. More that if you haven't built a taxonomy yet, then you are probably running before you can walk. Get the basics right and the rest will follow.