Copy Copy

December 9, 2010 7:50 AM

How sensitive are plagiarism checkers?

How many words in a row need to be identical in order for a plagiarism checker to identify them?

If a person takes a full paragraph and changes one word in the middle of it, would that cause a plagiarism checker to miss it?

posted by crazylegs to computers & internet (6 answers total)

If a person takes a full paragraph and changes one word in the middle of it, would that cause a plagiarism checker to miss it?

No.

http://en.wikipedia.org/wiki/Plagiarism_detection
posted by phunniemee at 7:53 AM on December 9, 2010 [1 favorite]

It sounds like your underestimating how complex these systems can be. You could catch up on things like regular expressions and Bayes, etc and that's probably more like what their underlying processes use.
posted by yeahyeahyeahwhoo at 9:00 AM on December 9, 2010

I have a class that runs all of our papers through those and it's based a "match" on two words. It always has a ton of false positives which makes me think it's pretty sensitive.
posted by biochemist at 1:56 PM on December 9, 2010

Anecdata: I had a group assignment this semester, and one of my classmates was completely useless and emailed me a draft of her section of the paper at 2am the day it was due. It was a really awful draft and I more or less rewrote it from scratch, but there were a few sentences I left basically intact (as in, the general theme of the sentence stayed the same, but I cleaned up the wording).

For whatever reason, she decided to submit her draft (just her draft, not the rest of the paper, not the cover sheet) to Turnitin without telling us. When I uploaded the final version to Turnitin later that day, it detected plagiarism, and I had an awesome time explaining to the lecturer why this had happened.

If a person takes a full paragraph and changes one word in the middle of it, would that cause a plagiarism checker to miss it? Nope.
posted by jaynewould at 3:50 PM on December 9, 2010

As a college professor who frequently battles plagiarism, I am my own plagiarism checker. It's hugely obvious when my students plagiarize. Basically, if you found it, I can find it too, and all it takes is Google, not a plagiarism checker.

If you're doing enough research that you find something that there is NO WAY for me to find the phrases you lifted online, then you might as well just take the little bit of extra effort to write your paper in your own words.
posted by Fuego at 6:23 PM on December 9, 2010 [2 favorites]

There are fast and relatively simple algorithms for finding the "edit distance" -- number of letters added + deleted -- between two texts. With enough data you could get a handle on what kind of edit distance is normal for different lengths of documents, and what signals plagiarism. I have no inside knowledge but I imagine this is how the plagiarism checkers work.
posted by miyabo at 9:06 PM on December 9, 2010