Support for users of CCG software closed 7-27-20

[Vivek Srikumar <vsrikum2 AT illinois.edu>]

Yes, the SRL demo takes the first role set. Picking the first sense of the verb is right 80-81 % of the time.

-- vivek  

On Thu, Jun 23, 2011 at 4:15 PM, Mark Sammons <mssammon AT illinois.edu> wrote:

The SRL demo just takes the first roleset (I believe).

Mark

---- Original message ----
>Date: Thu, 23 Jun 2011 14:08:24 -0700

>From: Steven Corman <steve.corman AT asu.edu>
>Subject: RE: [Illinois-ml-nlp-users] Need a user manual
>To: Mark Sammons <mssammon AT illinois.edu>, Geetali Dudhbhate
<gdudhbha AT asu.edu>

>Cc: "illinois-ml-nlp-users AT cs.uiuc.edu" <illinois-ml-nlp-users AT cs.uiuc.edu>

>
>Mark,
>
>When you say
>
>>     Our demo uses the PropBank frame files, presenting
>>     the most common
>>     set of labels for each identified verb.
>
>Do you just take the number 1 roleset for a verb, or does SRL select one that you
use to fill in the argument descriptions?  If the latter, is there a way to find out which it
selects?
>
>Thanks,
>
>Steve
>
>-----Original Message-----
>From: illinois-ml-nlp-users-bounces AT cs.uiuc.edu [mailto:illinois-ml-nlp-users-
bounces AT cs.uiuc.edu] On Behalf Of Mark Sammons
>Sent: Thursday, June 23, 2011 1:32 PM
>To: Geetali Dudhbhate
>Cc: illinois-ml-nlp-users AT cs.uiuc.edu
>Subject: Re: [Illinois-ml-nlp-users] Need a user manual
>
>Hi, Geetali.
>
>There is nothing distributed as part of SRL to automatically generate the more
verbose labels -- at least, not in the current release. This would be a problem because
our web demo uses the PropBank frame files, which are part of the copyrighted
distribution from LDC.
>
>The best I can suggest at present is, if you have the PropBank corpus, to implement
a postprocessing step that reads the SRL output, and finds the relevant frame file for
each labeled verb to look up the argument labels.
>
>Regards,
>
>Mark
>
>---- Original message ----
>>Date: Thu, 23 Jun 2011 13:28:05 -0700
>>From: Geetali Dudhbhate <gdudhbha AT asu.edu>
>>Subject: Re: [Illinois-ml-nlp-users] Need a user manual
>>To: Mark Sammons <mssammon AT illinois.edu>
>>Cc: rizzolo AT gmail.com, illinois-ml-nlp-users AT cs.uiuc.edu
>>
>>   Hi Mark,
>>
>>   Thanks for the link!!
>>
>>   Is there any way that I can make the output more
>>   verbose as mentioned in my initial query? Is there
>>   any instructional manual that maybe lists the
>>   options for various different types of output? Is
>>   there some input parameter which I have to provide
>>   to the code to make the output more verbose.
>>
>>   Thanks & Regards,
>>   Geetali.
>>
>>   On Thu, Jun 23, 2011 at 1:18 PM, Mark Sammons
>>   <mssammon AT illinois.edu> wrote:
>>
>>     Hi, Geetali.
>>
>>     The SRL was trained on the PropBank data; the
>>     labeling scheme has apparently changed slightly
>>     since the SRL training data was developed.
>>
>>     Here is a starting point for information about
>>     PropBank:
>>
>>     http://verbs.colorado.edu/~mpalmer/projects/ace.html
>>
>>     Our demo uses the PropBank frame files, presenting
>>     the most common
>>     set of labels for each identified verb.
>>
>>     Regards,
>>
>>     Mark
>>     ---- Original message ----
>>     >Date: Thu, 23 Jun 2011 12:29:55 -0600
>>     >From: Geetali Dudhbhate <gdudhbha AT asu.edu>
>>     >Subject: [Illinois-ml-nlp-users] Need a user
>>     manual
>>     >To: rizzolo AT gmail.com
>>     >Cc: illinois-ml-nlp-users AT cs.uiuc.edu
>>     >
>>     >   Hi,
>>     >
>>     >   I have managed to install the SRL package
>>     now. On
>>     >   giving input as "At the end of the day ,
>>     251.2
>>     >   million shares were traded ." , I get the
>>     output as
>>     >   :
>>     >
>>     >   (S1 (S (PP (IN At)
>>     >              (NP (NP (DT the)
>>     >                      (NN
>>     end))
>>     >                  (PP (IN of)
>>     >                      (NP (DT
>>     the)
>>     >                         
>>     (NN
>>     >   day)))))
>>     >          (, ,)
>>     >          (NP (QP (CD 251.2)
>>     >                  (CD million))
>>     >              (NNS shares))
>>     >          (VP (AUX were)
>>     >              (VP (VBN traded)))
>>     >          (. .)))
>>     >   At                   -   
>>     (AM-TMP*
>>     >   the                  -    *
>>     >   end                 -    *
>>     >   of                    -    *
>>     >   the                  -    *
>>     >   day                 -    *AM-TMP)
>>     >   ,                    
>>     -    *
>>     >   251.2              -    (A1*
>>     >   million             -    *
>>     >   shares            -    *A1)
>>     >   were              -    *
>>     >   traded          trade    (V*V)
>>     >   .                   
>>     -    *
>>     >
>>     >   If I run the same sentence on the SRL demo
>>     provided
>>     >   on
>>     >  
>>     http://cogcomp.cs.illinois.edu/demo/srl/results.php
>>     >   I also get some keywords like "temporal
>>     [AM-TMP]"
>>     >   and "thing being traded (away) [A1]". Is
>>     there any
>>     >   way in which I can make the output of the SRL
>>     >   package I have installed more verbose? Also
>>     is there
>>     >   some user manual to which I can refer, to
>>     better
>>     >   understand the meaning of the various tags
>>     used?
>>     >
>>     >   Thanks & Regards,
>>     >   Geetali Dudhbhate.
>>     >   MS-Computer Science.
>>     >   Arizona State University.
>>     >________________
>>     >_______________________________________________
>>     >illinois-ml-nlp-users mailing list
>>     >illinois-ml-nlp-users AT cs.uiuc.edu
>>     >http://lists.cs.uiuc.edu/mailman/listinfo/illinois-ml-nlp-users
>
>_______________________________________________
>illinois-ml-nlp-users mailing list
>illinois-ml-nlp-users AT cs.uiuc.edu
>http://lists.cs.uiuc.edu/mailman/listinfo/illinois-ml-nlp-users

_______________________________________________
illinois-ml-nlp-users mailing list
illinois-ml-nlp-users AT cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/illinois-ml-nlp-users