Sentiment, social media, and the leaders
Tonight is going to be an important test, not just of three political leaders, but of a branch of computer science called sentiment analysis.
This sets out to analyse language and try to draw conclusions about its "emotional" content.
A lot of different firms are going to try to use this technique to examine what people say online during the first prime-ministerial debate tonight, and to work out how positive or negative the language is about each of the three party leaders.
We've been talking to one firm , which specialises in what it calls "sentiment and text-analysis solutions", or "machine-readable news".
The people at Lexalytics say their technology has been used by the finance industry, where it helps firms to track corporate reputations and even make predictions about where share and commodity prices might be heading.
It's also been employed in American politics. The firm's CEO Mark Thompson told me the system, which can track thousands of online news sources, blog comments, tweets and other social-networking activity, had spotted the rise of a future president: "What we found in late 2007 was that this chap Obama was making a lot of noise online, and we couldn't work out why."
The company is now planning to track sentiment towards Messrs Brown, Cameron and Clegg during tonight's debate, and Mr Thompson is making a big claim: "We think during the debate we'll be able to follow what the world is thinking in real time."
We've asked the company to track Twitter activity, blogs and mainstream news before, during and after tonight's debate. We'll report back tomorrow with the results.
I'll also monitor a couple of other attempts to take the online temperature. , the site which tracks political tweets, will be producing a real-time sentiment ticker, as well as recording the volume of conversations and the most-tweeted terms.
And Facebook is inviting its users to "rate the debate", with a system which allows them to click on a dial to show positive or negative they feel about what they're hearing throughout the 90 minutes.
A word of caution - all of these techniques are relatively untried, so we won't be drawing any firm conclusions from their results. But they should help make watching the debate even more exciting.
I'm 
Comment number 1.
At 15th Apr 2010, barry white wrote:A high tech way to work out who will PM? I do worry when this type of thing happens, but lets see if if works, eh?
Complain about this comment (Comment number 1)
Comment number 2.
At 15th Apr 2010, BluesBerry wrote:2007, Obama was making a lot of noise online, and we couldn't work out why. Has Lexalytics ascertained the answer yet?
I’m still trying.
I have to admit I’m interested in your report tomorrow; this has increased my interest in watching the debate, just to learn how out of step I am with the rest of the world – or at least with the Tweeters and the Face-Book users.
Complain about this comment (Comment number 2)
Comment number 3.
At 16th Apr 2010, Briantist wrote:BluesBerry: What's with your hyphenation? It makes you look like you're using DOS 3.11.
Complain about this comment (Comment number 3)
Comment number 4.
At 16th Apr 2010, Monjo wrote:Blues-Berry, what is up with your hypenation? Face Book is clearly two words.
I suspect many now wish Obama had made less noise online and more noise giving change we could believe in.
Complain about this comment (Comment number 4)
Comment number 5.
At 16th Apr 2010, davednh wrote:There are a number of tools that showed the popularity of Obama on the rise in 2007 - search frequency, word counts, etc. - but the key problem is what seems to be missed in this article: "...Obama was making a lot of noise online, and we couldn't work out why." - that's the challenge - figuring out why. It will be interesting to see if Lexalytics not only sees popularity in the PM debate but moves forward in "finding out why". I am looking forward to their report.
Complain about this comment (Comment number 5)