Profile: TagCloud

Service: TagCloud

Launched: June 2005

Location: Bellwood, PA

Status: Corporate name is IonZoft

What is it?

TagCloud is a service that generates a “tagcloud” (see below) based on provided URLs or feeds. A tagcloud is basically a grouping of tags associated with content, from a single or multiple sources. Tagclouds are a visual tool – tags used more often are bigger and/or darker than less used tags, allowing a visual representation of relative tag use. Clicking on a tag takes the user to relevant content related to the tag – basically a list of content associated with such tag (ranking is usually by date, or freshness, but this is not a requirement.

TagCloud is a tool for creating tagclouds. A key fact to point out right at the start is that TagCloud uses keyword (text) analysis, not tag/category analysis, in creating the tag cloud. This is a point of contention (or at least discussion) around the service and worth noting as you read on.

Here’s a visual example of a tagcloud (remember that in a real tag cloud each item is clickable and links to relevant content):

The reason tagclouds are useful, instead of merely visually entertaining, is that they show in a very clear way the most popular tags and link directly to associated content. This will become increasingly important as the RSS aggregation and search applications collide in the near future (underlined because I am going to point back to this post in 2 years and say “I told you so”).

If you are mathematically inclined, I highly recommend reading Pietro Speroni’s three articles on tagclouds linked to below (see Links). There really is something to this other than a neat visual trick.

TagClouds is an easy to use service, and other than the contentious issue of using text/keyword analysis instead of category/tag analysis, the only problem is that they are extremely slow to generate. Since a key feature is adding these to your blog, the load time can be prohibitively long. An easy fix would be for Tagcloud to cache content periodically and present that cached (and slightly dated) content instead of re-generating the tag cloud on every html or xml call – I assume they are considering this. The speed issue is a killer and I would much prefer slightly old data instead of waiting 2-3 minutes for my blog to fully load in a browser.

Key Features:

– easy sign up
– easy to create a cloud based on a single feed or url
– easy to create a cloud based on a group of feeds
– import an opml file for easy creation of a massive cloud
– good tools for publishing clouds on websites
– dedicated URLs for each cloud, in html and xml format
– show any number of tags in the cloud, up to 250

As Pietro writes, a tag set or tagcloud for a single source of information is not very interesting. For instance, here is the tagcloud for Techcrunch only:

Link (long load time)

Compare this to the tagcloud for our entire opml file, with over 200 feeds of awesome blogs and other content (see list here):

Link (long load time)

It is a much more interesting result set. Remember, clicking on any tag gives a result set. Here is part of the result set after clicking on the “search” tag above:


Tagcloud uses Yahoo’s Content Analysis web service
and simple php to put the feed information into a database, analyze it with the Yahoo tool and generate the cloud:


Given their openness to creation of new applications using tagcloud, this can be an exciting and very useful service over time.

Hint: If aggregators generated a tag cloud for all of my unread content every day, it would be much easier for me to peruse posts that I am interested in. This can solve the overload problem we are all facing as we try to track more and more feeds daily (reading over 200 daily feeds takes me at least 2-3 hours per day right now). What I would like to see is a tag cloud based on tags and keywords.


John Herren (his original experiment here)


Pietro speroni on tagclouds theory in general: (doesn’t fully apply because PS is based on tagging, not keywords) #1 (good bibliography), #2, #3
Implementation Guide (to put on your website)
Recommended Web Tools
Ben Ramsey
Hans on Experience
Knowledge Jolt with Jack
Search engine
Marc Canter
Stephan Mosel
MicroContent Musings
Sudar Blogs
Tris Hussey
Stowe Boyd


Pete Freitag on how to make a tagcloud

Tags: , , , , , , , , ,