Tag clouds can be sorted in different ways: alphabetically, by frequency, semantically and randomly. Will McGugan prefers them sorted by frequency, because that makes them look better. In the comments people pointed out that it is easier to find items in a sorted list. Will responded he chooses form over function.
Navigation aid?
But what exactly is this function then? UI-patterns summarizes the problem a tag cloud is supposed to solve as:
The user wants to browse content by popularity or most elaborate topic
In short: a tag cloud is a navigation aid. The user wants to browse and is in need of information on popularity to decide where he wants to go next.
Fish-eye, birds-eye
This description of a tag cloud is valid if the tag cloud is considered from a fish-eye perspective. The user looks at it from his current location, has a clear idea where he wants to go and is presented with several ‘doors’ to choose from. In that case it really makes sense to order these doors in a predictable manner, e.g. alphabetically.
On the other hand, a tag cloud from a birds-eye perspective has a different purpose. The user does not want to know which particular options he has, but he wants an overview, for orientation. A problem with classic sitemaps is that they easily get cluttered for sites with many pages. A tagcloud solves this, by showing only the landmarks, or by clearly marking those landmarks.
Orientation aid
A tagcloud can be compared with a window display for a shop. It shows what kind of items the shop has to offer. But it is not a catalogue, where items are classified in such a way that someone who is looking for a particular item can find it quickly. Alphabetic ordering would only makes sense if a tag cloud were meant as a look-up table. But for this purpose there are other (better) tools available: hierarchies, search forms, indexes, etcetera. As an orientation aid frequency ordering makes more sense than alphabetic ordering, because with the former it is easier to find the ‘landmarks’.
Alternative sorting principles
There are many more possibilites when it comes to sorting. Semantic clustering sounds very promising, where items are placed together according to their semantic distance (closeness in meaning). Technically this is much more difficult than frequency sorting or alphabetic sorting, because it requires processing to a semantic level.
No Comments