backstage.bbc.co.uk

Use Our Stuff To Build Your Stuff

Prototypes

Who's in the News?

Description
This is similar to the wiki proxy, but inside out: first it extracts people, places and things from the latest news stories, (using the Lingua::EN::NamedEntity Perl module) and then tells you which news stories refer to them - the result: which people, places and organisations are making the BBC news right now.

Lingua::EN::NamedEntity is far from perfect, and that's pretty much the major weakness in this prototype.

  • 12 May 2005 01:51 PM

Comments  Post a comment

Very nice.

A clever addition might be for users to login and enter a list of names of people/places/organisations of interest and they get emailed whenever any person in their list is mentioned in the news.

  • 2.
  • On 16 May 2005 12:29 PM,
  • knc said:

Images need some filtering.

  • 3.
  • On 16 May 2005 04:05 PM,
  • nate said:

Images definately need filtering, though the 'Saddam Insane' photo is quite amusing.

Andrew: Surely that's something that should be done from subscribing to search results. (Email? How very 90s. Shouldn't you be using RSS for that now?)


knc: Hey, if you can come up with an open source programmatic image filtering system, I'll happily add it.

Oh, and before I forget: if anyone else wants to try this, the Java-based ANNIE named-entities extractor (free from http://www.gate.ac.uk/) might be a better way to go.

Post a comment




Remember Me?




style: lo-fi | hi-fi