Having just finished marking first-year essays about the Link Economy... I've been having a quick play with a tool called Evri, the new semantic network tool that's being used most notably by the Washington Post and The Times. If you have a look at the bottom of this and the three other most recent blog posts on the Centre for Journalism site, you'll get the idea. Evri searches through each post and uses fancy semantic web search methods to suggest key terms about which readers might want to find out more information. 

As you'll see, it's pretty good - although does throw up some oddities. Tim's Iran Votes posting, for example, throws up links to stories about Moussavi, but not Ahminedinejad. And the links about Paul Francis - following his posting about the NHS - rather disturbingly suggest a number of articles about Gary Glitter. I'd be interested to know what others think, and whether we should use it as a permanent feature of the site.



Am I being a total fruitloop or am I the only person simply getting their front page and no "content recommendations"? Or to get them, am I supposed to sign myself up?

Hmmm. Very strange. You shouldn't have to sign up - the recommendations should automatically appear. They do for me even if I'm not logged in to the site. Anybody else having this difficulty?

Ian Reeves is head of the Centre for Journalism

It's present here. It's present everywhere, really. Surprisingly.

It's Java, right? If so...?

It works in the newsroom, so the problem is my Mac. Dundundun.

On the other hand, I've just been having a little look at it - I like it. It's interesting.

p.s. Oopsie daisy. I finally updated my Mac. It's working now. Must remember to update it more often...

CfJ testing the new Evri semantic tool