Here are the slides & notes for a presentation I gave last night to the Sydney Sharepoint User Group on the topic of Sharepoint & Taxonomies. The presso is basically in 3 sections.
Slides 2-14: Introduction to Taxonomies
I started off by asking the audience how they might group wine together (or classify it) – the answers included colour, variety, vintage, region, bottle shape, sweetness. Then we had a look the way a wine shop orders them. The point here (apart from giving Brendan a bit of a plug) is that there are many ways to group things – and some of the most useful ones for users/consumers are not necessarily obvious from the object itself. Hierarchies & facets were then discussed via the systems of Dewey* (Dewey Decimal) and Ranganathan (Colon Classification) and some real-world examples.
The “we don’t need structure, we just need search” comment also got a mention – which resonated with a few people in the audience. The answer(s) to this include: i. taxonomies & metadata can make search better & ii. taxonomies & informtion mapping are about more than just findability. We ended that segment with Patrick‘s taxonomy map.
Slides 15-25: Taxonomies in Organisations
This section could be represented by a 3 x 3 matrix – who is involved in taxonomies vs what they are doing. I split the “who” into 3 broad groups:
- Experts – and here I mostly mean taxonomy experts but it could also be subject matter experts.
- Machines – language processing / semantic software (but this could also include process automation software as well).
- Users – general people who just do, y’know, stuff.
You need to involve all 3 groups but each has their strengthens & weaknesses. And then I tackle 3 broad activities:
- Building a taxonomy (or folksonomy or ontology).
- Applying terms to documents.
- Consuming – which in this situation means doing things with documents based on their metadata. This could as simple as someone searching & finding something or some fancy processing based on an ontology.
Slides 26-37: Sharepoint
Sharepoint’s basic methods of managing metadata are:
This is a good start but Sharepoint has three main deficits:
- It doesn’t handle hierarchical relationships between terms in lists well – it treats each list as though it is independent.
- Metadata can easily get caught in site “islands”.
- It doesn’t do any of the fancy machine classification.
A range of third-party vendors have arisen to meet these needs – each offering very different functionality at varying costs.
*I don’t know whether to be offended or impressed by Dewey’s classification of Australia with extra-terrestrial worlds.