Thursday, May 31, 2007

Hot or Not? Heat theme maps

Around the web these days you see lots of "heat theme" maps, pulling out the "hot" keywords in blocks of text (mainly blogs) but it can be anything (example). They take a block of text, and find all the words used most in it. Most of the time they use words from a specific database, but they don't have to: http://services.disruptiveproactivity.com/heattheme/

This sort of visualisation of words is also useful in others areas - analysing the output from questionnaires and other responses. Or large reports where you want to see what's talked about lots, and what isn't. So I connected the code from the mySociety panopticon (which tells us what people are saying about our sites) and attached the input to a HTML form, rather than the database. Which lets you put a load of text into a box and get a diagram of what's mentioned lots.

But we already have a service designed to make commenting on large reports easier, and which makes a text version available for doing things with. So we also feed the output of the text version into the input of the heat theme form, and find what the Transformational Government document spends most of the time talking about: like this.


Thanks to William for his help in testing it - and for giving me the idea in the first place.

Have fun.

2 Comments:

Anonymous Murph said...

I was pointing out the heat theme maps to some colleagues on our school governors forum and give them an example analysis to look at. The top item of the map is nbsp which I guess refers to the   entities which litter the site furniture.

Is there a way of excluding entities entirely from the theme maps?

13/6/07 06:54  
Blogger Sam Smith said...

I've knocked out nbsp and other common entities. I'll try and knock them out before the punctuation gets stripped but that will take a bit more time

13/6/07 08:59  

Post a Comment

Links to this post:

Create a Link

<< Home