I've never been to a Pargram meeting,
so I don't know if this is appropriate, because it is not about
XLE:
I'd like to have a group discussion about formulating a Pargram
reply to the Google universal part of speech tag set and the
Google universal dependency label set. They are gaining
momentum and most people in the ACL crowd accept it without
questioning. Linguists can ignore Google, but it's a little
scary for the field of language technologies. If we wait too
long, there will be too many resources in the Google format and it
will be too late to change it. (Although maybe it's not so bad
because Google based their dependency labels on Stanford
dependency, which is based on LFG.)
In fact, there are a few projects on Treebank interoperability
(HAMLETD from Prague, something from Reut Tsarfaty, Google
universal dependency set, and a relatively new project by me, Owen
Rambow, and Archna Bhatia). Pargram has been working on
interoperability for 20 years, so I would like to hear the Pargram
position.
Some sample questions:
1. HAMLETD did a survey of how coordinate structures were handled
in 29 dependency treebanks. They found that there were four or
five different ways, and then they picked one and converted all
the treebanks to that format. Sorry for skipping the first 19
Pargram meetings and not knowing this, but in Pargram is there a
one-size-fits-all treatment of coordinate structures, or is there
some variation across languages?
2. I'm worried that Google is not allowing enough
cross-linguistic variation on IOBJ and OBJ2. They might be
forcing languages to have grammatical relations that they don't
have.
--Lori
On 2/10/2014 5:57 PM, Kaplan, Ronald wrote:
If you haven’t already told us, please let us know whether you
will be coming to the meeting, and whether you would like to have
a presentation slot on the agenda or whether there is a particular
topic you would like to see discussed.
It’s now about 2 weeks away, so we need to get down to more
specific plans.
Thanks—
Ron
Begin forwarded message:
Subject:
February
Pargram meeting
Date: January 27, 2014 at
6:25:08 PM PST
Hi everyone,
As indicated earlier, we will be hosting the meeting at
Nuance in Sunnyvale, February 24-28. If you want to
look up the location on a map, the address is 1198 E.
Arques Avenue, at the intersection of Lawrence
Expressway and Central Expressway. The nearest train
station is the Lawrence station, about a 10 minute walk.
It is now time to develop the schedule. Following
tradition, our tentative proposal is to allocate blocks
of time roughly as follows:
Monday: Feature committee and parse comparison (Jani
has sent out the sentences)
Tuesday morning: Grammar reports
Tuesday afternoon: General session (proposal: a
tutorial/discussion on grammar debugging in XLE)
Wednesday morning: Papers/discussion
Wednesday afternoon: Symposium at CSLI/Stanford
celebrating Pargram’s 20th anniversary
Thursday: Papers/discussion
Friday morning: Planning and wish list, especially
about the new arrangements for XLE distribution and
maintenance.
Please let us know whether you plan to come (for local
logistics), and also let us know whether you would like
to present a paper, suggestion a topic for discussion,
or organize a more focused workshop session.
Best,
Annie and Ron
_______________________________________________
ParGram mailing list
ParGram@mailman.uni-konstanz.de
https://mailman.uni-konstanz.de/mailman/listinfo/pargram