"I'm researching about Faceted Navigation and Information Retrieval. I've been looking over the Internet for some articles/books/white papers about which is the best number of facets to use on a classification."
Interesting question, especially given the popularity of faceted search and taxonomy. The community discussed the topic, and a a few answers were provided by members.
"The answer is it depends on what you want to use the facets for, and how simple/predictable they are.
If you want to use facets for navigation, users typically start to get confused if you present them with more than three facets at a time (but can tolerate more if they are very predictable and simple).
If you're using facets to support other content management tasks or search, from the user's point of view you can have any number working in the background, but from the management point of view obviously the more you have, the more tricky it is to maintain them. The golden rule is don't develop a facet unless it's going to do some useful work for you."
My answer to the question was also the consultant standard "it depends".
Essentially, it depends on the context and the level of expertise of the user. For example, if you are in an e-commerce setting, and you are selling to the masses at the top level of your navigation, then less is definitely more. In terms of top level nav guidelines, most folks use the old 7+/- 2 rule from cognitive science (where 7 is the average number of things people can keep in their mind at any given time). Some sites get away with 10-12, but really, 6-7-8 is ideal in terms of scannability. As you scan across a number of choices, you have to evaluate them against each other, so having more than people can keep in mind at once means they have to go back and forth. User testing is the best way to gauge the right number.
Where it starts getting interesting is when you drill down in one facet and expose additional facets for refinement. Here the tolerance is lower in general, and only creeps up if you are an expert user with specialty needs. For example, if I'm an average shopper and I drill down into cameras on a website, then I might care to see facets for Brand, Price, Megapixels, Zoom... That's starting to push it in terms of number of facets. 3 or 4 is usually the most you want to show fully (as Patrick mentioned)... But you can show additional facets for those expert users (e.g. photography enthusiasts), and just collapse them so that they don't overwhelm everyone else. Only the hard core will ever go perusing in your lesser-known facets.
A similar pattern exists within the enterprise: you can usually get away with a handful of facets in a search refinement interface... Again, 3 or 4 - whatever fits above the fold typically. People won't dig too deep if they are not expert users.
So to recap, 3 or 4 is usually the sweet spot for refinements...
You can give more for expert users as long as you use UI tricks to avoid overwhelming folks
7 +/- 2 is a good rule of thumb for navigational categories.
If you’re looking for more on faceted taxonomy design, I wrote an article called “Designing for Faceted Search” that has some general tips and Seth and I cover that topic in a podcast.