3 ms·

Hey, I'm a TA at some CS courses, can this be used to find similar codebases? Maybe it useful to find similar programmers, find possible code copies and/or iden

by aaossa 9y ago

Hey, I'm a TA at some CS courses, can this be used to find similar codebases? Maybe it useful to find similar programmers, find possible code copies and/or identify the author of a snippet. Any ideas?

jacobwilliamroy 9y ago
Why are you interested in such applications?
- spc476 9y ago
  "TA" means "teaching assistant." I think the OP is asking about this to detect cheating and/or plagiarism.
dsfyu404ed 9y ago
There's only so many ways to do the toy problems that people get assigned. Students code winds up looking like their sources because they copy the control structures because there's only so many sane ways to do those problems without introducing fluff. At least this is what my friend's experience using plagiarism checking tools was.
- userbinator 9y ago
  because there's only so many sane ways to do those problems Indeed, that's been my experience too; if you assign something "Hello world" level, you're going to get a bunch of 100% correct solutions that look almost the same, and then the other... odd ones which are almost certainly incorrect in some way. It's only when you assign something far more nontrivial, and with many correct ways to do it, that you will get more diverse solutions; and even then, the chances of the correct solutions being very similar are high.
zaphos 9y ago
People often talk about using MOSS for this: https://theory.stanford.edu/~aiken/moss/ https://theory.stanford.edu/~aiken/moss/
hanbura 9y ago
Authorship attribution is the main application discussed in the paper, and there is some thought given to the possibility of detecting ghost writers (i.e. detecting if you let someone else write your code) I think it's promising, but to evaluate it against existing solutions and make it production ready is probably still a good chunk of work. For simple code similarity there are Moss, JPlag and Sherlock (linked in the paper )
sillysaurus3 9y ago
Your students are probably hiring people off Wyzant to do the problems for them. It's impossible to identify the author when they're a willing participant.

by aaossa 9y ago

jacobwilliamroy 9y ago

Why are you interested in such applications?

spc476 9y ago
"TA" means "teaching assistant." I think the OP is asking about this to detect cheating and/or plagiarism.

dsfyu404ed 9y ago

There's only so many ways to do the toy problems that people get assigned. Students code winds up looking like their sources because they copy the control structures because there's only so many sane ways to do those problems without introducing fluff. At least this is what my friend's experience using plagiarism checking tools was.

userbinator 9y ago
because there's only so many sane ways to do those problems Indeed, that's been my experience too; if you assign something "Hello world" level, you're going to get a bunch of 100% correct solutions that look almost the same, and then the other... odd ones which are almost certainly incorrect in some way. It's only when you assign something far more nontrivial, and with many correct ways to do it, that you will get more diverse solutions; and even then, the chances of the correct solutions being very similar are high.

zaphos 9y ago

People often talk about using MOSS for this: https://theory.stanford.edu/~aiken/moss/ https://theory.stanford.edu/~aiken/moss/

hanbura 9y ago

Authorship attribution is the main application discussed in the paper, and there is some thought given to the possibility of detecting ghost writers (i.e. detecting if you let someone else write your code) I think it's promising, but to evaluate it against existing solutions and make it production ready is probably still a good chunk of work. For simple code similarity there are Moss, JPlag and Sherlock (linked in the paper )

sillysaurus3 9y ago

Your students are probably hiring people off Wyzant to do the problems for them. It's impossible to identify the author when they're a willing participant.