What Shape is a Taxonomy?
One of the big problems about taxonomies (there are many) is that we keep getting kicked back to the biological definition of what it should look like and how it’s structured. In particular we are told a taxonomy needs to have a hierarchical tree structure following some very strict rules about how each branch relates to its peers but particularly to the concepts above it.
“Scientific” hierarchy rules include things like:
- Inclusiveness – the top category “includes” all subordinate categories in the tree (so the Indian caste system which might look like a social hierarchy does not qualify as a scientific hierarchy, because the Brahmana caste does not contain all subordinate castes)
- Relational consistency – the kind of relationship between each level in the hierarchy is exactly the same relationship (so “steering wheel is a part of a car, which is a kind of vehicle” would fail this requirement because the relationships change from part of to kind of as you move up the tree)
- Inheritance – subordinate categories in a hierarchy inherit all of the attributes of superordinate categories, which makes it easier to focus just on the differentiating attributes and makes hierarchical taxonomies very economical to use – by knowing which branch of the tree we are in we can already say a lot about anything we find in that branch (again, our Indian caste hierarchy fails this criterion)
- Mutual exclusivity – an entity can belong in one and only class which is why this strict sense of hierarchy is so attractive – it eliminates ambiguity completely (unfortunately, biological objects tend to follow these rules automatically in consequence of the laws of genetics but other manufactured objects and mental objects don’t).
Scientific hierarchies are very attractive in principle because of the way they eliminate ambiguity, enforce consistency, ensure rigourous observation and knowledge of the field, and identify relationship rules that produce predictability. Beautiful, you might think, for making knowledge content easily findable.
Think again. In practice, the world (and especially the world of organisational knowledge) is by no means as tidy or as well-defined as the scientific paradigm would suggest. Many of the categories we work with in normal life overlap with each other and sometimes compete. They rebel furiously against being pinned down.
In knowledge management we’re not dealing with well defined things with defined beginnings and endings (like birth and death) so much as documents and people containing mish-mashes of concepts, ideas and possibilities, at greatly varying levels of definition and clarity. These woolly, hard-to-pin-down-definitively knowledge agglomerates can rarely be described consistently by multiple people based on observation of physical characteristics. So what value is the predictability of a strict hierarchy? We’d never get all our knowledge and information assets to fulfil all those conditions of transitivity and inheritance for a start… maybe that would not be such a bad thing, empty knowledge repositories are very easy to maintain.
Moreover, in scientific taxonomies, our “objects” always sit at the end points of the hierarchy branches – all the levels above the end node where the organism sits are abstractions that help us manage our knowledge about those organisms. In a knowledge management taxonomy, our “knowledge objects” can sit at any level of the taxonomy from most general to most specific – as any librarian will tell you. They won’t sit neatly at the end nodes.
The truth is, hierarchies are useful only in certain specific circumstances; trees with more flexible rules about branching (so long as they are consistently applied) are more useful in other circumstances; and other structural forms such as facetted lists, matrices and concept maps are also valid and useful for particular contexts and purposes. Barbara Kwasnik wrote a wonderful article listing a variety of forms taxonomies can take, which has influenced my more detailed analysis in my forthcoming book. Part of the art of the taxonomist is in figuring out which form is most appropriate for the purpose at hand.
So if we can’t recognise a taxonomy by its shape, how then shall we know it? Read more on another day if you are still curious!
0 Comment so far
Commenting is not available in this weblog entry.Comment Guidelines: Basic XHTML is allowed (<strong>, <em>, <a>) Line breaks and paragraphs are automatically generated. URLs are automatically converted into links.
