Skip to Content.
Sympa Menu

illinois-ml-nlp-users - RE: [[Illinois-ml-nlp-users] ] Getting semantic roles

illinois-ml-nlp-users AT lists.cs.illinois.edu

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

List archive

RE: [[Illinois-ml-nlp-users] ] Getting semantic roles


Chronological Thread 
  • From: "Sammons, Mark" <mssammon AT illinois.edu>
  • To: Arvind Agarwal <arvind385 AT gmail.com>
  • Cc: "arvind385801 AT gmail.com" <arvind385801 AT gmail.com>, "illinois-ml-nlp-users AT lists.cs.illinois.edu" <illinois-ml-nlp-users AT lists.cs.illinois.edu>
  • Subject: RE: [[Illinois-ml-nlp-users] ] Getting semantic roles
  • Date: Tue, 30 Aug 2016 16:42:48 +0000
  • Accept-language: en-US

HI, Arvind. 

The version of SRL currently integrated with Curator does not provide the sense prediction.  I just released a new version of our standalone NLP pipeline that includes SRL, and this version will report lemma and sense number for each predicate. This is a significantly newer version that that in Curator; it will take a while before Curator can be updated with this version.

 

You can find the new pipeline here:

 

http://cogcomp.cs.illinois.edu/page/download_view/IllinoisPreprocessor

 

 

Regards,


Mark

 

From: Arvind Agarwal [mailto:arvind385 AT gmail.com]
Sent: Tuesday, August 30, 2016 9:54 AM
To: Sammons, Mark
Cc: arvind385801 AT gmail.com; illinois-ml-nlp-users AT lists.cs.illinois.edu
Subject: Re: [[Illinois-ml-nlp-users] ] Getting semantic roles

 

Hi,

Thanks for your reply.

 

Could you please provide me a bit more detail on this post-processing step. Did you mean writing my own code to extract role-specific information from XML files available in the propbank data; or is there a provision in SRL to do so? I can write an XML parser to get the role specific inform from XML files  however for that I need the sense. Currently I am not even getting the sense. 

 

The output I get is:

A0(saw, Mr. Smith)
A1(saw, the dog with a telescope)

 

How do I know that we are talking about see-0, and not see-02 or see-03.

 

Thanks once again for your help. Really appreciate it.

 


Best,

Dr. Arvind Agarwal, PhD

Research Scientist

 

 

On Tue, Aug 30, 2016 at 8:11 PM, Sammons, Mark <mssammon AT illinois.edu> wrote:

Hi, Arvind.

Right now, the only way to get the role-specific info is to add it in a post-processing step -- the SRL in Curator can only provide the coarse argument roles.

Regards,

Mark


-----Original Message-----
From: arvind385801 AT gmail.com [mailto:arvind385801 AT gmail.com]
Sent: Tuesday, August 30, 2016 2:55 AM
To: illinois-ml-nlp-users AT lists.cs.illinois.edu
Subject: [[Illinois-ml-nlp-users] ] Getting semantic roles

Hi,
I am trying to run the Curator's SRL annotator. For the sentence

# Mr. Smith saw the dog with a telescope.

I get the following output:

A0(saw, Mr. Smith)
A1(saw, the dog with a telescope)


I need some additional information such as

Predicate: See-01
A0 - Viewer
A1 - thing_viewed

I know that this information might be available from Propbank but it might ve difficult to parse all the frames and extract  the information. I was wondering if it directly available from the annotator.

Thanks for your help.

 




Archive powered by MHonArc 2.6.19.

Top of Page