[Air-l] turnitin issue

[burkx006 at umn.edu]

Hm, interesting.

I mentioned before that Lucas Introna has been investigating and writing 
about Turnitin. Since the matching algorithm is unknown to the users, he 
has been probing it as a "black box" by submitting systematically altered 
passages of text. He reports that certain kinds of word changes, and even 
single word changes at certain positions, will be flagged by the algorithm, 
while word changes at other positions cause the copied passage to pass 
unnoticed.

I thnk what troubles Introna (and I tend to agree) is that we don't know 
the reasons that some passages are flagged as possible plagiarism matches 
and some are not. Relying on the service essentially cedes the decisional 
criteria to an unknown programmer.

Since you deliberately "seeded" your exercise with errors, you knew what 
you expected to be flagged for the students to review. But based on 
Introna's reports, I'm uncertain whether the service would work as well, 
for example, in the exercise to have students assess their own papers for 
acceptable citation.

DLB


On Mar 9 2007, Marj Kibby wrote:

>In my first year cultural studies course I give students a turnitin
>report of a paper I wrote for the purpose that has a range of citation
>errors and a number of correctly cited quotes.
>
>In small groups they go through the report, decide which 'matches' are
>ok, and which are citation errors that need to be fixed. They add
>references, quotation marks, paraphrase, etc as required. 
>
>We discuss what the various groups have done, and they are encouraged to
>go through their own reports before final submission of the papers.
>
>Turnitin is set up to allow submission for a week or two before the due
>date, and for students to see their own reports immediately.
>
>I think where Turnitin is solely used as a detection tool, there is a
>tendency to look just at the percentage of matching text - but I haven't
>found that particularly useful. I think you need to examine every
>instance of matching text to see just why and how it matches.
>
>Marj
>
>
>
>
>
>Dr Marjorie Kibby, 
>Senior Lecturer in Communication & Culture
>Faculty of Education and Arts
>The University of Newcastle,  Callaghan NSW 2308 Australia
>Marj.Kibby at newcastle.edu.au
>+61 2 49216604
>>>> James Whyte <whyte.james at yahoo.com> 03/10/07 3:34 AM >>>
>
>In that vein, I am not sure that I understand Marjorie's claim that 
>Turnitin is useful for teaching referencing, since no one outside the 
>company knows the matching algorithm -- the criteria for text comparison
>
>are unknown, so it is hard for me to see what the students would learn. 
>Perhaps she could say more about that.
>
>
> _______________________________________________ The 
> air-l at listserv.aoir.org mailing list is provided by the Association of 
> Internet Researchers http://aoir.org Subscribe, change options or 
> unsubscribe at: http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org
>
>Join the Association of Internet Researchers: 
>http://www.aoir.org/
>

-- 
Dan L. Burk
Oppenheimer, Wolff & Donnelly Professor
University of Minnesota Law School
229 19th Avenue South
Minneapolis, MN  55455
**********************************
voice: 612-626-8726
fax: 612-625-2011
bits: burkx006 at umn.edu