Path: utzoo!lsuc!ncrcan!hcr!larry
From: larry@hcr.UUCP (Larry Philps)
Newsgroups: comp.edu
Subject: Re: looking for cheating detectors
Message-ID: <3655@hcr.UUCP>
Date: 26 Jul 88 13:08:32 GMT
References: <1403@sbcs.sunysb.edu>
Reply-To: larry@hcr.UUCP (Larry Philps)
Organization: HCR Corporation, Toronto
Lines: 25

In article <1403@sbcs.sunysb.edu>, paco@sbcs.sunysb.edu (Francisco J Romero) writes:
> 
>   I am looking for a program which detects similarities between/among C 
>   programs, and can be used to detect cheaters.  
>   Maybe
>   there are more sophisticated ways of detecting cheaters I am not aware of.
>   Any help will be appreciated.                                           

When I was at the University of Toronto, a rather ambitious Grad student wrote
a program that processed the parse tree of a compiled program in the Turing
language.

It used the hcr to generate the parse trees for all the programs in the
class, then compared each of those trees with each of the others and generated
a metric reflecting how similar they were.  The output was sorted in order of
decreasing metrics.

It was incredibly interesting to go looking for the listing of two progams that
had a 100% comparison metric, go through them line by line, and see the
industriousness of a student who had changed every variable name, all the
formatting, and all the comments, yet still had not changed the operation of a
single line of code.

As someone else in this discussion mentioned however, it is much more of a
problem to decide what to do with cheaters once you have caught them.