Corpora: Counting semantic propositions (was Relatve text length)

[Diego Molla]

There is no need to regard language as an ideal object to find semantic
propositions. You can try and device a robust approach that extracts
some sort of core semantic representation, this is feasible with the
current technology.

This is the central idea in answer extraction and question answering
systems like ExtrAns and AnswerFinder. These systems find the sentences
with highest predicate overlap with the predicates generated by the
question. And you guess right, this list of predicates is automatically
produced from any text sentence or question.

For more information you can have a look at:

http://www.ifi.unizh.ch/CL/extrans/
http://www.ics.mq.edu.au/~diego/answerfinder/

AnswerFinder's Web page is still very preliminary, I'll add more
information shortly.

Diego

Tadeusz Piotrowski wrote:

 > I know some people love semantic propositions etc., but for me we are
 > back again in the world of Platonic ideas. I like this discussion group
 > because language is usually not regarded here as an ideal object. I must
 > confess I find counting (calculating) ideal objects like semantic
 > propositions a bit difficult. I find it difficult both as a researcher
 > and as a practising translator, and I reach for my Quine to find peace
 > of mind.
 > Regards
 > Tadeusz Piotrowski
 >
 >
 >>-----Original Message-----
 >>From: owner-corpora at lists.uib.no
 >>[mailto:owner-corpora at lists.uib.no] On Behalf Of Alex Chengyu Fang
 >>Sent: Monday, April 29, 2002 5:34 PM
 >>To: Yorick Wilks
 >>Cc: ramesh at ccl.bham.ac.uk; corpora at hd.uib.no
 >>Subject: Re: Corpora: Relatve text length
 >>
 >>
 >>What I wanted to say is that there are different ways
 >>of measuring the relative length and that, if counts
 >>of characters, syllables and morphemes are used, you
 >>are likely to see differences between language pairs.
 >>If, however, semantic proposition is used as key,
 >>lanauges may not be so different as the number of
 >>propositions should be a near constant across
 >>multi-lingual texts that are mutual translations of
 >>each other.
 >>
 >>So, my simplistic view is that to see the differences,
 >>use characters, syllables and morphemes as
 >>measurements. To see similarities (the other
 >>direction), the number of semantic propositions can
 >>serve the purpose.
 >>
 >>Regards,
 >>
 >>Alex
 >>



--

      This message is intended for the addressee named and may
      contain confidential information.  If you are not the intended
      recipient, please delete it and notify the sender.  Views expressed
      in this message are those of the individual sender, and are not
      necessarily the views of Macquarie University.

---------------------------------------------------------------------
Diego MOLLA ALIOD                                 diego at ics.mq.edu.au
Department of Computing               http://www.ics.mq.edu.au/~diego
Macquarie University