Experiments With and Thoughts About Tag Clouds

We’ve been looking into using tag clouds as an alternative means of navigation for a long lists of links that defy easy categorization.

How to Organize Lots of Links?

Basically, we’ve got a number of directory pages with links to hundreds of articles provided by a content vendor. We’re trying to determine how best to organize these links.


We’ve done some high-level categorization, but to have further sub-categorization on each directory page would require having multiple links to the same article in different but similar sub-categories.

And, it’s not always obvious which sub-category a link should go in.
User testing also showed that it was difficult to name the sub-categories so it was really clear what links they contained.

They were still too high level to be really useful. However, we don’t have the time or resources to create additional levels of categories.

Alphabetical Listing

We could just list all the links alphabetically on each directory page, but that would make the page not easily scannable.

In addition, to do this properly we would need to rewrite many of the links to ensure that the each began with the appropriate word.

For example, “What to do in an emergency” would be listed in the W’s as it is written. Not very useful for someone looking for the word “emergency”. We would have to rewrite it to something like “Emergencies: What to do in them”.

While this is obviously a good idea, there are numerous maintenance and resource issues that prevent us from being able to do this.

Enter Tag Clouds

Because of this issue, we’ve been wondering about using tag clouds as an alternative navigation option at the top of each directory page.

Categorization is handled by extracting the keywords from each link so we don’t have to even worry about which categories a link best fits into. This should enable us to get around our maintenance issues.

To that end I’ve been trying out ZoomClouds, nice service which will create a tag cloud from an RSS feed.

It’s very easy to customize and style to your own specific design. An alternative is TagCloud but their site is currently undergoing an overhaul and so I wasn’t able to try them out.

What No RSS Feed?

Our main problem is that the content does not generate an RSS feed — once it goes live it’s going to be pretty static. So, how to generate the tag cloud?

Well, fortunately I found Cloud JS which uses JavaScript to generate the tag cloud and can read an XML file.

Even better, the developer provides a tool to scrape your content and generate an XML file automatically so it looks like my problems may be solved.

Do Tag Clouds Even Work?

Of course, whether the tag cloud concept will work for our audience is still completely unproven, so a considerable amount of user testing still lies ahead. Zeldman, for one, is certainly not a fan.

On that note, is it me or has the concept of using tag clouds for navigation not really caught on?

I don’t seem to see them around that much on the web (except for that Web 2.0 crowd) — if at all on any sites that have a non-technical audience. Will tag clouds ever break out into the mass market?

Further Reading About Tag Clouds

4 thoughts to “Experiments With and Thoughts About Tag Clouds”

  1. There’s a lot of potential with tag clouds that hasn’t yet been explored yet (I have some ideas on this I’m experimenting with) but in the meantime, the names (tag cloud, weighted list) are cumbersome for non-techy people. The first time I saw tag clouds on sites like delicious, I didn’t know what they were called and I didn’t need to know – I just saw it as a list of topics yelling out to me at different volumes. I like the link you’ve listed in the related links section but I’m really not sure about the label – perhaps “topics” or “categories” might have been more helpful.

  2. Rachel – I like the concept of tag clouds; it’s just that there can be many issue regarding their implementation.
    This is especially true when you base them off of user-generated tags. There can be many variations of a tag — e.g., web_20, web2.0, web20, web-2.0, etc.
    Unless your can group these related tags into a single link within a tag cloud, the benefit of the weighted list will be greatly diminished — you’re splitting the vote, so to speak.
    Of course, if you have some sort of controlled classification then the weighted list concept works much better.

  3. Tag clouds looks like an innovative concept… but the only problem i see is that the categorization works in an automated way… it would be difficult to adapt it to known information.

Comments are closed.