Davy, I really liked your prototype. I've worked for a while on something called sentiment detection, which is to do with determining whether an author likes or dislikes some topic/product. What you are doing has a lot of similarities, although there is less emotional language in news - which is generally meant to be objective, though I think that more and more evaluative language is slipping in there.
You probably don't want to let on too much about how you are doing things, but I would be interested to know a little more about the type of approach you are taking. Is it lexical/grammatical, or is it some form of machine learning/statistical classification based on some form of training? I guess a naive approach would use phrases that indicated topics which were generally good or bad. However, that approach would be easily thwarted by things like 'bombing suspect arrested' - which one might think of as good news. Anyway, I'd love to hear more about your approach. Matt Hurst Senior Research Scientist Intelliseek, Inc BlogPulse.com On 7/8/05, Davy Mitchell <[EMAIL PROTECTED]> wrote: > Hi Folks, > > Just signed up last night so HI! > > I've just published an Alpha stage static page of my project Mood News > An overview of the days news from the BBC, auto-classified as Good, > Bad or Neutral. > > http://www.latedecember.com/sites/moodnews/ > > Please provide feedback :-) Yeah the formating needs improved... > > Thanks, > Davy Mitchell > > > - > Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please > visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. > - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html.
