I was delighted to be introduced by Mark Davey to Leala Abbott on Monday. Leala is a smart and accomplished digital asset management consultant from the Metropolitan Museum of Art in New York and we were discussing how difficult it is to explain what we do. I told her about how I describe “the evolution of classification” to people and she asked me to write it up here. So, this is my first blog post “by commission”.
In the beginning there was the word, then words (and eventually sentences).
Then people realised words could be very useful when they were grouped into lists (and eventually controlled vocabularies, keyword lists, tag lists, and folksonomies).
But then the lists started to get a bit long and unwieldy, so people broke them up into sections, or categories, and lo and behold – the first taxonomy.
People then realised you could join related taxonomies together for richer information structuring and they made faceted taxonomies, labelling different aspects of a concept in the different facets.
Then people noticed that if you specified and defined the relationships between the facets (or terms and concepts), you could do useful things with those relationships too, which becomes especially powerful when using computers to analyse content, and so ontologies were devised.
Here is a very simple example of how these different KO systems work:
I need some fruit – I think in words – apples, pears, bananas. Already I have a shopping list and that serves its purpose as a reminder to me of things to buy (I don’t need to build a fruit ontology expressing the relationships between apples and other foodstuffs, for example).
When I get to the shop, I want to find my way around. The shop has handy signs – a big one says “Fresh fruit”, so I know which section of the shop to head for. When I get there, a smaller sign says “Apples” and even smaller ones tell me the different types of apples (Gala, Braeburn, Granny Smith…). The shop signs form a simple taxonomy, which is very useful for helping me find my way around.
When I get home, I want to know how to cook apple pie, so I get my recipe book, but I’m not sure whether to look under “Apples” or “Pies”. Luckily, the index includes Apples: Pies, Puddings and Desserts as well as Pies, Puddings and Desserts: Apples. The book’s index has used a faceted taxonomy, so I can find the recipe in either place, whichever one I look in first.
After dinner, I wonder about the history of apple pies, so I go online to a website about apples, where a lot of content about apple pies has been structured using ontologies. I then can search the site for “apple pie” and get suggestions for lots of articles related to apples and pies that I can browse through, based on the ideas that the people who built the ontology have linked together. For example, if the article date has been included, I could also ask more complex questions such as “give me all the articles on apple pies written before 1910″, and if the author’s nationality has been included, I could ask for all the articles on apple pies written before 1910 by US authors.
People often ask me if a taxonomy is better than a controlled vocabulary, or if an ontology is the best of all, but the question doesn’t make sense out of context – it really depends what you are trying to do. Ontologies are the most complex and sophisticated KO classification tools we have at the moment, but when I just want a few things from the shop, it’s a good old fashioned list every time.