Hi, Rob - this is neat, though not entirely sure that it's working
entirely as you might want...
http://muddyboots.rattleresearch.com/cgi-bin/mb.cgi?action=page&id=701
<http://muddyboots.rattleresearch.com/cgi-bin/mb.cgi?action=page&id=701>
...a page about "The Sun" (and the "News of the World") has lots of
links off to the NASA website - presumably because of the use of the
word "Sun"...
Nice, though - and something to think about.
Hi James,
Thanks for this, it highlights one of the challenges we face when trying
to find correct contextual meaning where ambiguity exists, we haven't
got it right in all cases yet :)
I thought I'd work it through and highlight areas that could be
improved. The initial story has been categorised as being related to
the following tags (via the yahoo term extraction service) :
(http://muddyboots.rattleresearch.com/cgi-bin/mb.cgi?action=view&id=701)
* media ownership
* editorial control
* ownership laws
* communications committee
* independent board
* evening newspapers
* evidence <http://en.wikipedia.org/wiki/Evidence>
* news corporation <http://en.wikipedia.org/wiki/News_Corporation>
* chairman <http://en.wikipedia.org/wiki/Chair_%28official%29>
* mr <http://en.wikipedia.org/wiki/MR>
* house of lords <http://en.wikipedia.org/wiki/House_of_Lords>
* news of the world <http://en.wikipedia.org/wiki/News_of_the_World>
* mr murdoch
* parliamentary committee <http://en.wikipedia.org/wiki/Committee>
* murdoch <http://en.wikipedia.org/wiki/Murdoch>
* fox news <http://en.wikipedia.org/wiki/Fox_News_Channel>
* sky news <http://en.wikipedia.org/wiki/Sky_News>
* sun <http://en.wikipedia.org/wiki/Sun_%28disambiguation%29>
* news station <http://en.wikipedia.org/wiki/News_station>
* rupert murdoch <http://en.wikipedia.org/wiki/Rupert_Murdoch>
The obvious problem with this is the "sun" tag, it is an ambiguous term
that has many meanings, as evidenced at :
http://en.wikipedia.org/wiki/Sun_(disambiguation)
Currently we only follow the links off these disambiguation pages to
gather external links, however if we were to improve our usage of the
disambiguation pages we could cut down on these false positives (in fact
that's top of the list of the things we'd like to experiment with).
The other problem here is that we display inks if they have any matches
in del.icio.us with the story tags listed above. We should probably put
some metrics around the minimum number of tags a story must match to be
a recommended link, in this case that would have meant we wouldn't have
recommended the 'planetary' sun links if we had a minimum match of 2 tags.
Thanks for the feedback !
-
Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please
visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html.
Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/
