[backstage] BBC Touch

[Chris Riley]

Hi all,

I've been hanging around here for a while, but today I've been thrust into the limelight.  Ian has quite cleverly found my latest mashup using BBC data and has already posted it as a prototype to the backstage website ( http://backstage.bbc.co.uk/prototypes/archives/2006/10/how_in_touch_is.html ), and has asked me to mail the list with some background, so here goes.  (I think Jem mentioned it in an earlier email too.)

BBC Touch http://cgriley.com/bbctouch/ takes data from the BBC News RSS feed and compares it with data from the (sort of hidden) popular new stories XML feed to get a % of how "in touch" the BBC news editorial staff are with the public and what we are actually reading.  The about page I've already written explains it a bit better that I have just now http://cgriley.com/bbctouch/about/

Why did I do it?  I don't have any desire to highlight any hidden agendas the BBC's editorial staff might have (although I guess it can), but more from an interest in how "in touch" are the BBC with what the public actually reads and cares about compared to what they think we do.  To help identify subjects we might care about more than others all the headlines I get I put through the Yahoo Content Analysis API to get some useful keywords (subjects).  On the web page you'll see subjects they want us to read about vs. what we're actually reading about for the past 24 hours, and past 2 weeks.

Hopefully over time we'll be able to identify trends in the subjects areas the BBC push but we never read, or vice versa, and use the % as a measure for how in touch the BBC get at particular points in time.  People with a better understanding of this will no doubt have better ways to investigate this data, but I hope this is a useful tool in the analysis.  In particular I think its useful for highlighting issues the public care more about. For instance a couple of says ago whilst Pakistan was the headline, most of us were reading the climate change story.

How have I done it?  I explain this in the about page, but essentially I poll the feeds every hour and store them, along with any Yahoo extracted subjects, in a SQL Server database.  The touch page then pulls this data out, and works out the touch value based on this data.

As for the visual representation, I thought about using a tag cloud for the subjects, but found the decreasing size of text in a list a simpler and cleaner approach (and Jem seems to like it, so that's good. ;o) )

Planned features - I already intend to add a page to this to show the highest and lowest touch values achieved, and the headlines / subjects that achieved them.  However if you want to suggest things I could do with the site, other views on the data, or just have some general feedback then please do reply to this mail or email me using the email address on the about page.

I hope this has inspired you to take another look at the feeds we already have access to and play with them in new ways.

Thanks,
Chris Riley