Re: [nukkad] An experiment in grouping news articles

["Harshal Chhaya" <>]

aryaholmes wrote:
> how does one subscribe to this?

One doesn't. One has to visit the website to see the grouped news.



Anantha Narayanan wrote:
> This is a beautiful venture. Can I know how these news items are
> grouped? Also which news sites are you taking into consideration?

Headlines that have similar words are put in the single group. For
example, all headlines with 'Pakistan' are in one group and those with
'blast' are in another. This sometimes causes unrelated headlines to
be grouped together - for e.g. news articles about the mine blast and
the court verdict in the Mumbai blasts case are put in a single group.
The software that does the semantic analysis is not smart enough to
understand the bigger picture and to know that these two blast-related
stories are not really related.

The news sources being currently used are Rediff, India news from
Yahoo! News and the BBC. I am also using the Grapevine newsletter as a
news source but I realised yesterday that a bug causes only old
newsletters to be used. Will fix this soon.

If you check out the site a few times over a few days, you may see
some other problems - either with how the stories are displayed or how
they are grouped or with something else. If you do, please let me
know.

- Harshal


----------------------------------------------------------------------------
So many gods, so many creeds, So many paths that wind and wind, While 
just the art of being kind is all the sad world needs. 
-Ella Wheeler Wilcox, poet (1850-1919)
----------------------------------------------------------------------------


------------------------------------------------------------------------------
To join/leave, use the form at: http://www.mumbai-central.com/nukkad/#options
This list is archived at: http://www.mumbai-central.com/nukkad/archive.html