Tag clouds as a heuristic

I have been fascinated by the concept of the tag cloud since I encountered it for the first time a few years ago on flickr and del.icio.us. I’m not always too sure what to make of them, but my interest in using the tag cloud as a form of knowledge representation was renewed at the recent ELI Conference.

During a session presented by George Siemens and Cyprien Lomas, they offered some interesting perspectives on using web-based tools to assist with data visualization. One of the tools was, Many Eyes, which provides the opportunity for unique visualization of data. One of the options in ManyEyes is the generation of a tag cloud from large amounts of text. I had come across ManyEyes some time ago and thought it was pretty interesting, but at the time I just didn’t have a strong sense of what I might use it for. As happens all too frequently, I let it get buried under a host of other tasks and other ideas demanding my attention.

During the session George Siemens commented that he will occasionally – when he is bored – take the last several months of posts from his blog and dump them into ManyEyes to create a tag cloud. Someone in the audience immediately picked up on this and suggested that we take the last several years of the Horizon Report and generate tag clouds for the sake of comparison. This idea really seemed to resonate with people in the session, and the back channel communication being conducted on Twitter immediately began to light up with this idea.

Picking up on this idea after the conference, Chris Lott generated tag clouds of the most recent Horizon Report, as well as for past reports from 2004 – 2007, and posted them to his blog. Interesting stuff…

This practice of generating tag clouds for knowledge representation also appeared last week following President Bush’s most recent state of the Union address, one version of which can be found here. Both of these examples suggest an interest in using tag clouds as a form of sense making that is gaining in popularity.

This recent activity and buzz surrounding tag clouds has increased my interest in their use as a form of knowledge representation. Sometimes I find that when I look at tag clouds of data sets I get a flash of insight that leads to some realization, something that helps me analyze underlying meaning and sub-text. And sometimes I look at tag clouds and see…well….just clouds. Data haze. However, I am thinking increasingly that there is something a tag cloud reveals in a way not otherwise possible. In this way I think that the creation of tag clouds represent a replicable method for directing inquiry and attention when we are engaged in learning or problem solving…in other words a heuristic.

I am a huge of using technology tools in ways that help us to do things that would otherwise not be possible…or otherwise so time consuming and tedious as to discourage a particular practice. Many Eyes, for its ease of use in generating tag clouds from large amounts of text, permits a level of analysis that would otherwise not be possible. Taking hundreds of pages of text and representing the frequency of key words can be accomplished in mere seconds. There is also another web-based tool, a bit less sophisticated than Many Eyes, called TagCrowd that permits similar tag cloud generation using text, as well as just entering a URL for a website…so perhaps some interesting potential here as well. Ultimately, I think this kind of practice opens the door for us to ask some interesting questions and perhaps lead to inquiry that might otherwise remain unexamined.

So, I am continuing to think about practical ways this might be employed in educational contexts to support learning and inquiry. Building on the basic blog / report idea from George Siemens, I think the following practices would also be of interest:

  • Tag clouds for individual and class sets of student papers / essays.
  • Tag clouds for speeches and lectures.
  • Tag clouds for analyzing the content of websites.
  • Tag clouds of classic pieces of literature.
  • Tag clouds generated from set of stories covering the same news event.

Just to name a few…

The other thing I wonder is, if we engage in these kinds of practices in the classroom, what are the kinds of questions we should be asking students to wrestle with? how do they interpret tag clouds? Are they of value in supporting learning? Inquiry?

In any event, I’m sure I’ll be doing my fair share of dumping data into Many Eyes and seeing what kinds of patterns and questions emerge. I remain excited to learn about other creative uses for the tag cloud that are bound to emerge, but one I’m thinking about is a web-based application that would create a tag cloud from selected podcasts (transcribed text content) where the tags are also links to the list of podcasts addressing that concept. Anybody know of something like that?

1 Response to “Tag clouds as a heuristic”


  • Jeff,

    Your post makes me think of the recent WASC review that my previous school went through and how the reviewers combed through information to see if we were moving forward on the goals we said we were. They really could have used some of the tools you mention here. I also think about Understanding by Design and looking backwards from assessments to instruction and content for connections. It would seem to me that we could try to run student text based assessments through Many Eyes, etc. to see how well our students reached the Essential Understandings that we hoped they would learn.

Leave a Reply