[developers] seeking ERG and PET help

[Paul Haley]

Greetings,

 

I am looking for several points of assistance with the use of the ERG with
PET.  Please accept my apology in advance if this is less than appropriate
for the list.  I trepidatiously welcome any correspondence telling me "where
to go"!

 

First, we could use help in tokenization, particularly with regard to
unknown words and MWE.  We are, in effect, looking for the behavior of the
on-line demo, albeit with PET.   We have an SWT user interface that
transmits to PET as a web service, so we would prefer a Java solution.

 

Second, we have done a lot of work to map ERG types/sorts/predicates to an
OWL ontology, but there is more to do.  For example, we want the predicates
of quantifiers, connectives, and prepositions rigorously classified.  This
is challenging for non-ERG authors for several reasons, as you are probably
aware.  (Even the nominal sorts can seem confusing when viewed purely
semantically.)  If there are existing ontologies that have good coverage or
if someone who understands the critical aspects of the ERG non-terminals so
as to assist with or review such an ontology, we are interested.

 

FYI, the source text is XHTML from textbooks and sites such as Wikipedia,
especially with regard to cellular biology and chemistry and, to a lesser
extent, physics and mathematics. This motivates us to want even better
handling of affixation, compounds, and various lexical forms, such as
chemicals with sub- and super-scripts, too.  .  In time, the content will
become broadly biomedical.

 

I would be happy to share details on the project by request.  And to all
involved, I would like to express my appreciation for the excellent work in
PET and the ERG.  We hope that what results will reflect well on your
efforts and yield some useful contributions.

 

Thank you,

Paul

 

Paul Haley

Automata, Inc.

(412) 716-6420

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.delph-in.net/archives/developers/attachments/20120604/efd906c0/attachment.html>