title generation

title generation

Post by grpmang » Thu, 06 Jan 2005 16:45:53


hi
i have decided to implement a title generator in natural language
processing which when given an input file lists all possible titles
that are most appropriate for it. this has to be completed within a
period of 2 months.
could you please help me as to where to start and how to go about it.
which programming language would be the most apt for this task? is
java fine?
please send replies to my inbox:
XXXX@XXXXX.COM
as i myself dont know where all i'm posting this message.

thank you for your time
sincerely,
najla
1st yr. MTech cse
National Institute of Technology
India

[ comp.ai is moderated. To submit, just post and be patient, or if ]
[ that fails mail your article to < XXXX@XXXXX.COM >, and ]
[ ask your news administrator to fix the problems with your system. ]
 
 
 

title generation

Post by kirt » Sat, 08 Jan 2005 15:23:28

The programming language itself isn't of paramount concern just yet -
figure out how you want to do this. Probably you'll want to consider
estimating similarity between text using Bayesian methods (consider the
Rainbow Bayesian tools) and/or the cosine similarity metric (or a
related metric). With only two months probably you want to run some
tests using these methods pretty soon. With more time you could think
about combining the results of the two methods. So as far as that goes
you don't need symbolic processing until you start generating the title
itself. Now for that, you don't necessarily need a lot of NLP
processing per se: titles are summaries in effect - the shortest and
most descriptive or most eye catching summary that you can generate.
So check out what the summarization people have to say about this
problem. If you can predict the domain, you can probably generate a
grammar for title generation. Even if you can't predict the domain,
you can extract some rules for title generation although for that
you'll probably need some kind of ontology to keep your titles safe
from something like: Dog flies spacecraft around the world (possibly
extracted from a story about the Russian dog in space Laika, for
example).

BTW - I didn't just invent this approach. I did this as part of a
project four years ago.

Kirt Undercoffer

[ comp.ai is moderated. To submit, just post and be patient, or if ]
[ that fails mail your article to < XXXX@XXXXX.COM >, and ]
[ ask your news administrator to fix the problems with your system. ]