[Air-l] turnitin issue

[elw at stderr.org]

> A few other notes to consider: Turnitin does not store the actual paper. 
> They store a hash of the paper, weakening the argument that IP is being 
> violated.


[A *hash*?  Really, come on.  A whole-document hash, certainly not, given 
their output and use.  A hash of paragraphs or sentences?  Maybe - but 
that gets us closer to being able to reconstruct the actual text, or at 
least assess similarity.]

If I were building an online plagiarism detection service, using very well 
understood information retrieval methods - term-document matrices, 
document vectors, and the like - I would find it fairly difficult NOT to 
store the student's work in a re-constitutable form.

--elijah