shakespeare on taxonomy
Shakespeare, explaining why a certain level of granularity is of value in a taxonomy. From Macbeth:
1st Murderer: We are men, my lord.
Macbeth: Aye, in the catalogue ye go for men; as hounds, and greyhounds, mongrels, spaniels, curs, shoughs, water-rugs, and demi-wolves are clept all by the name of dogs: the valued file distinguishes the swift, the slow, the subtle, the housekeeper, the hunter, every one according to the gift which bounteous nature hath in him clos’d; whereby he does receive particular addition, from the bill that writes them all alike: and so of men.
An occasional challenge in designing a taxonomy, whether for back-end architecture, user permissions, catalog searchability, or any complex schema, is deciding how much detail to track. Too little, and like Macbeth, you find yourself unable to distinguish the good from the bad, the useful from the unwanted. Too much, and the relevant is hidden by the irrelevant, discouraging users.
How to know what to do? Like everything else, start with the user.
user-driven granularity
One of the simplest means of determining granularity is to leave it up to the user. There are several approaches, any or all of which can be helpful.
- Conduct an audit of search terms, internal and external. Navigation flow should also be compared to path analysis. Observing how users look for information provides valuable insight into their internal categorization. The user's mental model probably operates as an ideal level of relevant granularity.
- Grant users the freedom to develop their own folksonomy via tagging. Many content management systems allow this with minimal effort. For large-scale enterprise endeavors, platforms such as Endeca can meet a broad variety of producer-driven and user-driven needs. User-generated metadata (UGM) is typically highly relevant.
- If possible, perform concept testing. As always, design to test for purpose (as opposed to stimulus-response testing), and give the users every opportunity to do instead of say. Good techniques include structured brainstorming and user-driven card sorts. Insight into the user's categories is your goal.
standards-driven granularity
User-driven research is not much help when you're developing a taxonomy from the ground up, with a built-in need for shared elements. So, turn to those who have already solved the problem for their own users.
- Is there an industry standard? Sometimes it's simple. Library databases, for example, have a specific minimum of elements all items must be cataloged against, as well as optional metadata. If your product has an established standard taxonomy, chances are your users will expect it and you should incorporate it. This is not a recommendation of slavish adherence to tradition—rather, tradition can be a springboard for you, and a clue into the level of granularity your product will require.
- If there is no industry standard, look to information management for inspiration. For example, the Dublin Core Element Set provides a highly useful metadata minimum standard.
refining granularity
Once you've fleshed out a basic schema, testing and observation will help refine it to the appropriate level of granularity. Don't fit it all into one menu. Too much detail, and users will be overwhelmed by the options and abandon your site.
Instead, parcel out information into appropriate locations:
- A global navigation for key areas., common to the entire site.
- Anchor links within a section or page.
- A search interface, either simple or advanced, or both. An advanced system can be a good way of introducing your user to key metadata.
- Tagging.
This allows depth of granularity within the appropriate context, so that unlike Macbeth, your users can find exactly what they need, when they need it.
August 2009
More in Taxonomy
More in UXtraordinary
