Hello,
I'm in the process of moving the WordNet::QueryData mailing list to
Google. Adding large batches of e-mail addresses to a Google group
requires approval, so you may not be able to see your subscription
immediately. I'm told this process will take 24-48 hours.
Also, some subscribers to the Yahoo group have their Gmail accounts
set up to block the adding of their email addresses to Gmail groups.
If you do not receive confirmation of being added to the Gmail group
in 2-3 days, and you wish to be subscribed, you will need to manually
add yourself.
Here's the new mailing list web page:
http://groups.google.com/group/wn-perl/
Jason
Thanks for your reply. Yeah I also found the same just now using the interfaces of WordNet 2.0.
Best
Debanjan
PS - Can anyone suggest is there any measure to find out similarity between a "region" relation, such as between america and world_trade_center or Afghanistan and al-qaeda.
It seems that "arrive" in Sense 1 doesn't have any hypernym { as in WordNet 2.0 , and i assume its the same in 2.1 }.
<snip O/p of "wn arrive -hypev " >
Sense 1 arrive, get, come
Sense 2 arrive, make it, get in, go far => succeed, win, come through, bring home the bacon, deliver the goods </snip>
HTH, Cheers, maheedhar.
On 21/10/05, Debanjan <dghosh5@...> wrote: > > HI: > I am using Query Data 1.38 for hypernym calculation. Basically I am > interested in sense#1 and so forth. This module is working pretty fine but I > got stucked with few examples. > Suppose I use , > $res = join(", ",$wn->querySense("arrive#v#1", hype) ); > > This does not give me any result. Any clue ? > > thanks in advance > Debanjan > > > ----- Original Message ----- > From: Jason Rennie > To: wn-perl@yahoogroups.com > Sent: Thursday, May 19, 2005 11:58 AM > Subject: [wn-perl] QueryData 1.38 > > is out! > > http://people.csail.mit.edu/jrennie/WordNet/ > > It simply fixes the bug that Beata noticed. The change worked as > expected and didn't break any tests, so I figured there was little > reason to wait. Send mail to this list if you notice any problems. > > Thanks, > > Jason > > > WordNet::QueryData web page: > http://people.csail.mit.edu/~jrennie/WordNet/ > > > > > > WordNet::QueryData web page: > http://people.csail.mit.edu/~jrennie/WordNet/ > > > > > > ________________________________ > YAHOO! GROUPS LINKS > > > Visit your group "wn-perl" on the web. > > To unsubscribe from this group, send an email to: > wn-perl-unsubscribe@yahoogroups.com > > Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service. > > ________________________________ >
It seems that "arrive" in Sense 1 doesn't have any hypernym { as in
WordNet 2.0 , and i assume its the same in 2.1 }.
<snip O/p of "wn arrive -hypev " >
Sense 1
arrive, get, come
Sense 2
arrive, make it, get in, go far
=> succeed, win, come through, bring home the bacon, deliver the goods
</snip>
HTH,
Cheers,
maheedhar.
On 21/10/05, Debanjan <dghosh5@...> wrote:
>
> HI:
> I am using Query Data 1.38 for hypernym calculation. Basically I am
> interested in sense#1 and so forth. This module is working pretty fine but I
> got stucked with few examples.
> Suppose I use ,
> $res = join(", ",$wn->querySense("arrive#v#1", hype) );
>
> This does not give me any result. Any clue ?
>
> thanks in advance
> Debanjan
>
>
> ----- Original Message -----
> From: Jason Rennie
> To: wn-perl@yahoogroups.com
> Sent: Thursday, May 19, 2005 11:58 AM
> Subject: [wn-perl] QueryData 1.38
>
> is out!
>
> http://people.csail.mit.edu/jrennie/WordNet/
>
> It simply fixes the bug that Beata noticed. The change worked as
> expected and didn't break any tests, so I figured there was little
> reason to wait. Send mail to this list if you notice any problems.
>
> Thanks,
>
> Jason
>
>
> WordNet::QueryData web page:
> http://people.csail.mit.edu/~jrennie/WordNet/
>
>
>
>
>
> WordNet::QueryData web page:
> http://people.csail.mit.edu/~jrennie/WordNet/
>
>
>
>
>
> ________________________________
> YAHOO! GROUPS LINKS
>
>
> Visit your group "wn-perl" on the web.
>
> To unsubscribe from this group, send an email to:
> wn-perl-unsubscribe@yahoogroups.com
>
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
>
> ________________________________
>
I am using Query Data 1.38 for hypernym calculation. Basically I am interested in sense#1 and so forth. This module is working pretty fine but I got stucked with few examples.
It simply fixes the bug that Beata noticed. The change worked as expected and didn't break any tests, so I figured there was little reason to wait. Send mail to this list if you notice any problems.
You've got it right, Ted. 1.39 goes with WN 2.1. Use 1.38
with WN 1.7.1 or 2.0. Use 1.26 with WN 1.6 or 1.7. This is
all detailed on the QueryData web page under "Download."
Thanks for the update! I really appreciate your efforts to put Windows
installation instructions together on your blog. This is an area where we
have limited experience (actually, no experience, to be really honest
about it) so your help and that of other Windows users is truly
appreciated.
I will go ahead and also forward this to the wn-perl@yahoogroups.com list,
since it sounds like it might bear more upon WordNet::QueryData than it
does WordNet::Similarity (although I find it strange that QueryData works
ok on its own, but then fails with Similarity). In any case,
WordNet::QueryData and WordNet::Similarity are two different projects,
although Similarity certainly depends on QueryData.
BTW, I notice from your blog that you are using WordNet 2.0 with QueryData
1.39. I *think* those might be incompatible - I *think* 1.39 only supports
2.1, and for 2.0 you should back off to QueryData 1.38. This is what can
be confirmed or denied via wn-perl.
Cordially,
Ted
On Sun, 25 Sep 2005, rschurchill_uk wrote:
> I had a few issues getting Wordnet::Similarity installed on Windows
> under Activstate Perl, due to what is probably, a makefile issue in
> Wordnet::QueryData. The windows version of Wordnet uses different
> filenames in the /dict directory than it's UNIX counterpart.
> Wordnet::QueryData works fine on it's own but when trying to make
> Wordnet::Similarity, path errors occur.
>
> I fixed the problem but I haven't delved any deeper to figure out why
> the wrong filenames were in Wordnet::QueryData. I've published an
> installation guide for Windows users on my blog.
>
> http://spaces.msn.com/members/outside-world/
>
>
> Richard Churchill
>
>
--
Ted Pedersen
http://www.d.umn.edu/~tpederse
As far as I can tell, it works with WordNet 2.1. Note that I'm
not supporting older version of WordNet going forward. The web
page still has links to old version of the module that should work with
old versions of WN.
Note that validForms is buggy (and has been buggy---I don't know if it
ever worked "correctly"). Documentation has been updated to
reflect this fact.
As Ben mentioned, WordNet 2.1 moved many hypernym and hyponym relations
to "instance of" (inst) and "has instance" (hasi) relations.
QueryData no longer relies on the lexnames file that used to come with WordNet.
I removed a lot of old, deprecated code with this release.
- lexnames: unintentional omission. But, the WordNet C library (and thus, also, the browser) hasn't used the file for several versions. The equivalent data is compiled in. It hasn't changed since version 1.6, IIRC,
which makes it pretty static. Randee just posted it to our FTP server, since another user asked for it. Available at: ftp://ftp.cogsci.princeton.edu/wordnet/2.1/lexnames
or http://wordnet.princeton.edu/2.1/lexnames
Okay, I bit the bullet and hand-coded this into QueryData. :) Thanks for the URL.
A few of the tests don't work; hopefully those will be easy to fix. Shouldn't be too long 'till QD works with WN 2.1...
Hi Jason --
- lexnames: unintentional omission. But, the WordNet C library (and thus,
also, the browser) hasn't used the file for several versions. The
equivalent data is compiled in. It hasn't changed since version 1.6, IIRC,
which makes it pretty static.
Randee just posted it to our FTP server, since another user asked for it.
Available at:
ftp://ftp.cogsci.princeton.edu/wordnet/2.1/lexnames
or
http://wordnet.princeton.edu/2.1/lexnames
- The new relations are `sense' relations.
- My WN Perl lib is currently just internal to the lab. Half because it
doesn't do much error checking. Half because it's not well documented.
-- Ben Haskell
-- WordNet Technical Support Specialist
On Wed, 14 Sep 2005, Jason Rennie wrote:
> Many thanks for the patch. But, still some loose-ends I need to tie up
> before releasing a new version:
>
> - What happened to the lexnames file? Is it no longer part of WordNet? Is
> there a replacement of some sort?
> - Are the new relations, inst & hasi, `sense' relations or `word' relations?
>
> Thanks,
>
> Jason
>
Many thanks for the patch. But, still some loose-ends I need to tie up before releasing a new version:
- What happened to the lexnames file? Is it no longer part of WordNet? Is there a replacement of some sort? - Are the new relations, inst & hasi, `sense' relations or `word' relations?
Almost forgot:
Full changes @ http://wordnet.princeton.edu/2.1/CHANGES
Only other one that looks pertinent to the db structure is that we now use
the same filenames on both Win + Unix.
-- Ben
On Tue, 13 Sep 2005, Jason Rennie wrote:
> QueryData does not work with WordNet 2.1. I haven't had time to figure out
> what was wrong & make the necessary changes. If anyone can summarize the
> changes made to the database format, that might help.
>
> Jason
>
AFAIK, the only change we made was the addition of 'instances'.
Instances are specialized hyponyms. e.g. "Ronald Reagan" is no longer a
hyponym, but an instance, of U.S. President.
Pointer symbols added are: '@i' and '~i'.
In my own WN Perl lib, I've added these as 'inst' and 'hasi' (for
"instance" and "has instance", respectively). I also added 'hypes' (combo
of '@' and '@i') and 'hypos' (combo of '~' and '~i'), because it's still
useful to include instances in the hyp(?:o|er)nym hierarchies.
Untested patch against WN::QD 1.38 at:
http://wordnet.princeton.edu/~ben/wnqd.1.38.patch
-- Ben Haskell
-- WordNet Technical Support Specialist
On Tue, 13 Sep 2005, Jason Rennie wrote:
> QueryData does not work with WordNet 2.1. I haven't had time to figure out
> what was wrong & make the necessary changes. If anyone can summarize the
> changes made to the database format, that might help.
>
> Jason
>
QueryData does not work with WordNet 2.1. I haven't had time
to figure out what was wrong & make the necessary changes. If
anyone can summarize the changes made to the database format, that
might help.
Is WordNet-QueryData ready for use with WordNet 2.1? For some reason I think it is not, although I don't recall why I think that. :) Has someone tried to use WordNet-QueryData with this version of WordNet (2.1) as yet? If not, I will do so in the near future (on Solaris or Linux).
Cordially, Ted
------------------------ Yahoo! Groups Sponsor --------------------~--> Get fast access to your favorite Yahoo! Groups. Make Yahoo! your home page
http://us.click.yahoo.com/dpRU5A/wUILAA/yQLSAA/wGfwlB/TM --------------------------------------------------------------------~->
Greetings QueryData folk,
Is WordNet-QueryData ready for use with WordNet 2.1? For some
reason I think it is not, although I don't recall why I think
that. :) Has someone tried to use WordNet-QueryData with this
version of WordNet (2.1) as yet? If not, I will do so in the
near future (on Solaris or Linux).
Cordially,
Ted
I think if you use Perl in Windows system, you just need to copy the
QueryData.pm file to the perl lib director.
--
Carl (Wanxiang Che)
On 7/24/05, Garrett Heifrin <garrett_heifrin@...> wrote:
>
> Hi,
>
> I installed NMake, and when I try Makefile.pl, I get the following
> error message. Is it that QueryData is not compatible with Perl
> 5.8.6? If not, should I get an earlier version of Perl, and, if so,
> which is most compatible?
>
> C:\WordNet-QueryData-1.38>perl Makefile.pl
> Perl lib version (v5.8.6) doesn't match executable version (v5.8.0) at
> C:\Perl\lib\Config.pm line 32.
>
> Thanks,
> Garrett
>
>
>
>
>
>
>
>
> WordNet::QueryData web page: http://people.csail.mit.edu/~jrennie/WordNet/
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
Hi Garrett --
It looks as though you have two versions of Perl installed, and your
v5.8.0 perl is listed before your v5.8.6 perl in your PATH variable. Or at
least, it's the one being found. (Could be an aliasing issue: perl
pointing to perl5.8.0 or something similar.)
I performed the following test on our Solaris machine here, which still
has a v5.6.0 perl installed:
# set environment variable to intentionally find the wrong Config.pm
$ export PERL5LIB=/usr/csl/perl/lib/5.8.0/sun4-solaris-thread-multi
$ perl5.6.0 -MConfig
Perl lib version (v5.8.0) doesn't match executable version (v5.6.0) at
/usr/csl/perl/lib/5.8.0/sun4-solaris-thread-multi/Config.pm line 21.
So, not WN::QD's fault.
Best of luck.
-- Ben Haskell
-- WordNet Technical Support Specialist
On Sun, 24 Jul 2005, Garrett Heifrin wrote:
>
> Hi,
>
> I installed NMake, and when I try Makefile.pl, I get the following
> error message. Is it that QueryData is not compatible with Perl
> 5.8.6? If not, should I get an earlier version of Perl, and, if so,
> which is most compatible?
>
> C:\WordNet-QueryData-1.38>perl Makefile.pl
> Perl lib version (v5.8.6) doesn't match executable version (v5.8.0) at
> C:\Perl\lib\Config.pm line 32.
>
> Thanks,
> Garrett
Hi,
I installed NMake, and when I try Makefile.pl, I get the following
error message. Is it that QueryData is not compatible with Perl
5.8.6? If not, should I get an earlier version of Perl, and, if so,
which is most compatible?
C:\WordNet-QueryData-1.38>perl Makefile.pl
Perl lib version (v5.8.6) doesn't match executable version (v5.8.0) at
C:\Perl\lib\Config.pm line 32.
Thanks,
Garrett
I have installed WordNet 2.1 to C:\Program Files\WordNet\2.1 and set
WNHOME to C:\Program Files\WordNet\2.1. See below
Tue 07/05/2005 14:22:42 C:\>
set | find /i "wn"
WNHOME=C:\Program Files\WordNet\2.1
When I attempt to install I get the following error when I try to
nmake test. There seems to be a difference in the directory stuctures
between WordNet 2.0 and WordNet 2.1.
Tue 07/05/2005 14:24:47 L:\perl\WordNet\QueryData\
WordNet-QueryData-1.38>
perl Makefile.PL
Writing Makefile for WordNet::QueryData
Tue 07/05/2005 14:25:02 L:\perl\WordNet\QueryData\
WordNet-QueryData-1.38>
nmake
Microsoft (R) Program Maintenance Utility Version 1.50
Copyright (c) Microsoft Corp 1988-94. All rights reserved.
cp QueryData.pm blib\lib\WordNet\QueryData.pm
Tue 07/05/2005 14:25:27 L:\perl\WordNet\QueryData\
WordNet-QueryData-1.38>
nmake test
Microsoft (R) Program Maintenance Utility Version 1.50
Copyright (c) Microsoft Corp 1988-94. All rights reserved.
C:\Perl\bin\perl.exe "-Iblib\lib" "-Iblib\arch" test.pl
v1.6: 1..36
v1.7: 1..38
v1.7.1: 1..37
v2.0: 1..54
ok 1
Loading index files. This may take a while...
Not able to open C:\Program Files\WordNet\2.1/dict/lexnames or
C:\Program Files\WordNet\2.1\dict\lexnames: No such file
or directory at blib\lib/WordNet/QueryData.pm line 299.
NMAKE : fatal error U1077: 'C:\WINDOWS\system32\cmd.exe' : return code
'0x2'
Stop.
When I do a dir of C:\Program Files\WordNet\2.1\dict\ there is no
lexnames file to be found. If I search for lexnames*.* I do get the
following, but I don't know what I should move around.
Tue 07/05/2005 14:41:08 L:\perl\WordNet\QueryData\WordNet-QueryData-1.38>
dir "C:\Program Files\WordNet\2.1\lexnames*.*" /s /b
C:\Program Files\WordNet\2.1\doc\html\lexnames.5WN.html
C:\Program Files\WordNet\2.1\doc\pdf\lexnames.5.pdf
C:\Program Files\WordNet\2.1\doc\ps\lexnames.5.ps
Thanks,
Nat Davis
I see. Thanks, Ben.
Beata
--- In wn-perl@yahoogroups.com, Ben Haskell <ben@c...> wrote:
> Some nouns appear in WordNet in the plural if that is how the sense is
> usually found.
> "years" == "old age", "years" == "long time" in this case.
>
> Other examples:
> "spirits" == alcoholic beverages
> "dealings" == relations
> "amends" == reparation
> "hoops" == basketball
>
> -- Ben Haskell
> -- WordNet Technical Support Specialist
>
> On Wed, 25 May 2005, beata_b_bk wrote:
>
> > --- In wn-perl@yahoogroups.com, Jason Rennie <jrennie@g...> wrote:
> > > So, the fix will require a non-trivial amount of restructuring
of the
> > > QueryData. It might be a few days or weeks before I find the
time to
> > > work on this.
> > >
> > > It appears that the first word returned by validForms will always be
> > > valid. So, as a short-term fix, I'd suggest that if you're using
> > > validForms, you only use the first of the words returned.
> > >
> > > Jason
> >
> > HI, Jason,
> >
> > Is there any reason for validForms to return the original, pluralized
> > form of a noun? I.e. years#n -> years#n year#n ?
> >
> > I am currently taking last element from noun list and first element in
> > verb list as my base form defaults.
> >
> > Cheers,
> >
> > Beata
> >
> >
> >
> >
> >
> > WordNet::QueryData web page:
http://people.csail.mit.edu/~jrennie/WordNet/
> >
> >
> >
> >
> >
________________________________________________________________________________\
__________________________________________________
> > Yahoo! Groups Links
> > * To visit your group on the web, go to:
> > http://groups.yahoo.com/group/wn-perl/
> >
> > * To unsubscribe from this group, send an email to:
> > wn-perl-unsubscribe@yahoogroups.com
> >
> > * Your use of Yahoo! Groups is subject to the Yahoo! Terms of
Service.
> >
> >
Some nouns appear in WordNet in the plural if that is how the sense is
usually found.
"years" == "old age", "years" == "long time" in this case.
Other examples:
"spirits" == alcoholic beverages
"dealings" == relations
"amends" == reparation
"hoops" == basketball
-- Ben Haskell
-- WordNet Technical Support Specialist
On Wed, 25 May 2005, beata_b_bk wrote:
> --- In wn-perl@yahoogroups.com, Jason Rennie <jrennie@g...> wrote:
> > So, the fix will require a non-trivial amount of restructuring of the
> > QueryData. It might be a few days or weeks before I find the time to
> > work on this.
> >
> > It appears that the first word returned by validForms will always be
> > valid. So, as a short-term fix, I'd suggest that if you're using
> > validForms, you only use the first of the words returned.
> >
> > Jason
>
> HI, Jason,
>
> Is there any reason for validForms to return the original, pluralized
> form of a noun? I.e. years#n -> years#n year#n ?
>
> I am currently taking last element from noun list and first element in
> verb list as my base form defaults.
>
> Cheers,
>
> Beata
>
>
>
>
>
> WordNet::QueryData web page: http://people.csail.mit.edu/~jrennie/WordNet/
>
>
>
>
>
________________________________________________________________________________\
__________________________________________________
> Yahoo! Groups Links
> * To visit your group on the web, go to:
> http://groups.yahoo.com/group/wn-perl/
>
> * To unsubscribe from this group, send an email to:
> wn-perl-unsubscribe@yahoogroups.com
>
> * Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
>
>
--- In wn-perl@yahoogroups.com, Jason Rennie <jrennie@g...> wrote:
> So, the fix will require a non-trivial amount of restructuring of the
> QueryData. It might be a few days or weeks before I find the time to
> work on this.
>
> It appears that the first word returned by validForms will always be
> valid. So, as a short-term fix, I'd suggest that if you're using
> validForms, you only use the first of the words returned.
>
> Jason
HI, Jason,
Is there any reason for validForms to return the original, pluralized
form of a noun? I.e. years#n -> years#n year#n ?
I am currently taking last element from noun list and first element in
verb list as my base form defaults.
Cheers,
Beata
So, the fix will require a non-trivial amount of restructuring of the
QueryData. It might be a few days or weeks before I find the time to
work on this.
It appears that the first word returned by validForms will always be
valid. So, as a short-term fix, I'd suggest that if you're using
validForms, you only use the first of the words returned.
Jason
Yes, the "rules of detachment" are prioritized. First match with a
corresponding word in WordNet gets returned.
(Array definitions below are from morph.c in the WordNet C library)
Outline of morphstring:
Return entry from exception list if it exists.
Return nothing for adverbs. (only check exception list)
if (NOUN) {
if (string ends with "ful") { strip the "ful" and remember to add it
later }
elsif (string ends with "ss" or is <= 2 characters long) {
return nothing
}
Then the endings that are tested are in the sufx array:
static char *sufx[] ={
/* Noun suffixes */
"s", "ses", "xes", "zes", "ches", "shes", "men", "ies",
/* Verb suffixes */
"s", "ies", "es", "es", "ed", "ed", "ing", "ing",
/* Adjective suffixes */
"er", "est", "er", "est"
};
If a word ends with such an ending, that ending is stripped and
replaced with the corresponding entry in the addr array:
static char *addr[] ={
/* Noun endings */
"", "s", "x", "z", "ch", "sh", "man", "y",
/* Verb endings */
"", "y", "e", "", "e", "", "e", "",
/* Adjective endings */
"", "", "e", "e"
};
-- Ben Haskell
-- WordNet Technical Support Specialist
On Tue, 24 May 2005, Jason Rennie wrote:
> So, as far as I understand the morphy man page:
>
> http://wordnet.princeton.edu/man/morphy.7WN
>
> QueryData is doing the "right" thing. There are no entries in
> verb.exc for hates, dotes, tapes, dines, so QueryData should return
> hat, dot, tap, din as valid forms of the verbs.
>
> Of course, some of the behavior of WordNet may not be completely
> documented, or it may be that the "rules of detachment" are
> prioritiezed---only the first match is valid. I don't know what the
> truth is. If you or anyone else knows or can find out what is going
> on, I'd be happy to update the QueryData code.
>
> Jason
>
>
> WordNet::QueryData web page: http://people.csail.mit.edu/~jrennie/WordNet/
>
>
>
>
>
________________________________________________________________________________\
__________________________________________________
> Yahoo! Groups Links
> * To visit your group on the web, go to:
> http://groups.yahoo.com/group/wn-perl/
>
> * To unsubscribe from this group, send an email to:
> wn-perl-unsubscribe@yahoogroups.com
>
> * Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
>
>
So, as far as I understand the morphy man page:
http://wordnet.princeton.edu/man/morphy.7WN
QueryData is doing the "right" thing. There are no entries in
verb.exc for hates, dotes, tapes, dines, so QueryData should return
hat, dot, tap, din as valid forms of the verbs.
Of course, some of the behavior of WordNet may not be completely
documented, or it may be that the "rules of detachment" are
prioritiezed---only the first match is valid. I don't know what the
truth is. If you or anyone else knows or can find out what is going
on, I'd be happy to update the QueryData code.
Jason
Hi, Jason and everybody,
I wonder if the following is the intended behaviour:
query with validForms(hating#v)
returns
hate#v hat#v
Now, there happens to be a verb "hat" (put on or wear a hat); however,
hating cannot be its -ing form - should be hatting, as in patting,
hitting, etc.
This creates a problem for me, since I am defaulting to the last
element for the base form. This idea derives from the noun pattern:
years#n -> years#n year#n
where the first member is the actual full form, and not the best guess
for the base form.
WordNet online produces the page for "hate" when queried on "hating".
It could be that the "hate" version just happens to be first in line...
Similar problems surface up for:
dotes#v -> dote#v dot#v
tapes#v -> tape#v tap#v
dines#v -> dine#v din#v
Any ideas as to how to go handle this?
Thank you,
Beata