[FoRK] My talk at Google about predicting bugs
Jim Whitehead <
ejw at soe.ucsc.edu
> on >
Fri Mar 17 17:02:11 PST 2006
I gave a talk at Google last week about recent work in my lab about
bug prediction. Google records all talks, and, what do you know, they
also make them available via Google video. Click below to watch me,
and my student Sung Kim.
> Click to watch: Predicting bugs in code changes using SCM information
> Description: Google TechTalks March 8, 2006 Jim whitehead Jim
> Whitehead is an Assistant Professor of Computer Science at the
> University of California, Santa Cruz. He has recently been
> developing a new degree program on computer gaming, the BS in
> Computer Game Engineering. Jim received his PhD in Information and
> Computer Science from UC Irvine, in 2000 Abstract: Almost all
> software contains undiscovered bugs, ones that have not yet been
> exposed by testing or by users. Wouldn't it be nice if there was a
> way to know the location of these bugs? This talk presents two
> approaches for predicting the location of bugs. The bug cache
> contains 10% of the files in a software project.
Perhaps our most interesting results are the ones given in the last
half of the talk. After analyzing the history of a software project,
we're able to use machine learning techniques to predict whether a
new change contains a bug, or is clean. We can do this with ~75%
accuracy (recall is lower).
This means you could commit a change to an SCM system, and based on
this commit, the system can come back and say, "Dear Mr/Mrs/Ms
Programmer. We are 75% sure that you have just committed a bug. We
regret to inform you that we can provide no guidance on the location
of the bug. Still, we feel it would be worth your while to inspect
your change carefully."
If you're interested, we can send you the papers we've written on the
methods we describe in the talk.
More information about the FoRK