[LON-CAPA-users] Plagiarism check in grading interface...

[Raymond Batchelor]

The problems I am grading are two-part problems.  The first part is machine graded radiobuttonresponse and the SECOND part is the essay for which I want the plagiarism check.

>From what my TA tells me, I infer that the plagiarism check WORKED for them PRIOR to grading the students, but afterward it did not.
i.e. She could get %similarity reported for multiply-selected students, beforehand, but NOT after grading was completed.

I 'believed' that I saw a different similarity report for the same student for the same resource (i.e. a match to a different student's submission).  However I could easily have been mistaken, because I had tried the check so many times, on so many resources, for numerous students and from different servers.  Heh.

So, I would assume you are correct about that part of it and I probably got confused (unless I can reproduce an incidence of this possible phenomenon).

Thanks for the patch.

Cheers,
Ray



----- Original Message -----
From: "Stuart Raeburn" <raeburn at msu.edu>
To: lon-capa-users at mail.lon-capa.org
Sent: Thursday, 9 August, 2012 22:54:53
Subject: Re: [LON-CAPA-users] Plagiarism check in grading interface...

Hi,

> I am puzzled about the plagiarism check.

Thank you for reporting this.

> I previously had observed that at least some kind of check in   
> grading/viewing interface worked for multiply selected students in a  
>  course.

Yes, it looks as though checking for multiply selected students worked  
in LON-CAPA 2.4 and older, but not thereafter, as a change made while  
addressing bug 5363 (http://bugs.loncapa.org/show_bug.cgi?id=5363)  
apparently restricted this to one student selected each time.

Anyway the resolution of this issue is discussed in my comment #5  
added to bug #5712 today. The required change has been included in the  
forthcoming LON-CAPA 2.11.0 release.

The bug report ( http://bugs.loncapa.org/show_bug.cgi?id=5712 ) also  
discusses how to address this issue for institutions running the  
current (2.10.1) release.

> Also, it returns different results  for that one person.

By different results do you mean the percentage similarity, and/or the  
identity of the submitter with a similar submission above threshold  
would change between grading passes?

I would be surprised to see that, given that the code iterates over  
all the submissions recorded for the particular resource, and reports  
the one with the highest similarity.

Does the resource have a single essayresponse item, or are there  
multiple essayresponse items? The mechanism used for archiving  
submissions for later similarity checking was not designed to  
accommodate multiple essayresponses in a single resource.

I would be curious if you still see different results for one  
particular student when you make different grading passes, when using  
2.11.0 (which includes the change mentioned to support multiply  
selected students).

The testdrive cluster:

https://testdrive.loncapa.org

(not a part of the main LON-CAPA network) is currently running a  
2.11.0 release candidate, if you have an opportunity to test that out.


Thanks,

Stuart Raeburn
LON-CAPA Academic Consortium


Quoting Raymond Batchelor <batchelo at sfu.ca>:

> I am puzzled about the plagiarism check.
> I previously had observed that at least some kind of check in   
> grading/viewing interface worked for multiply selected students in a  
>  course.
> However, in the last couple of days, this is no longer working for   
> me.  I can ONLY get a check for the first student (alphbetically)   
> checked and NO others.
> Also, it returns different results  for that one person.
>
> I am beside myself trying to figure out what might have changed.
> What could be interfering with the "% similarity" output?
>
> Any ideas?
>
> Thanks!
>
> --
> Ray

_______________________________________________
LON-CAPA-users mailing list
LON-CAPA-users at mail.lon-capa.org
http://mail.lon-capa.org/mailman/listinfo/lon-capa-users