I'm perturbed that anyone thinks that it's possible to " detect deception automatically", as Norah Dunbar apparently does.
I *hope* everyone on this list accepts that TurnItIn, etc., merely detect textual similarities, not plagiarism. Similarly, any automated analysis of citations should only be thought of as a way of identifying grounds for concern, not as providing evidence of misconduct.
Julian
-----Original Message-----
From: Plagiarism [mailto:[log in to unmask]] On Behalf Of Dr. Mike Reddy
Sent: 19 July 2016 07:50
To: [log in to unmask]
Subject: Re: A New Scientist article related question
Just realised I can't easily add attachments to this list. However, here are the pertinent details:
Article: "Got a secret? Watch what you say"
Two paragraphs before the last read:
'Markowitz found this when he looked at research papers that had been retracted due to scientific misconduct. These tended to have more references than papers describing legitimate results.
This kind of research could pave the way for systems that detect deception automatically, says Norah Dunbar at the University of California, Santa Barbara. At the moment, we are still trying to figure out if there are recognisable linguistic patterns at all. "But I think the long-term goal for a lot of people in this area is to be predictive," she says.'
It shouldn't be hard for iParadigms to see if there is a correlation between originality scores and the size (relative or absolute) of the bibliography for entries in their database. Now that would be interesting!
Mike
> That would be great if you could scan it. Maybe it's early morning lack of concentration but I found the issue and couldn't find the article.
>
> Thanks
> Kevin
*************************************************************************
You are subscribed to the JISC Plagiarism mailing list. To Unsubscribe, change your subscription options, or access list archives, visit http://www.jiscmail.ac.uk/lists/PLAGIARISM.html
*************************************************************************
*************************************************************************
You are subscribed to the JISC Plagiarism mailing list. To Unsubscribe, change
your subscription options, or access list archives, visit
http://www.jiscmail.ac.uk/lists/PLAGIARISM.html
*************************************************************************
|