Bug 68061 - Add a script to figure out area of expertise of contributors
Summary: Add a script to figure out area of expertise of contributors
Status: NEW
Alias: None
Product: WebKit
Classification: Unclassified
Component: WebKit Website (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Nobody
URL:
Keywords:
: 71957 (view as bug list)
Depends on: 72090 72243
Blocks:
  Show dependency treegraph
 
Reported: 2011-09-14 00:40 PDT by Ryosuke Niwa
Modified: 2013-02-10 19:19 PST (History)
6 users (show)

See Also:


Attachments
Classifier, proof of concept 1 (5.51 KB, text/plain)
2011-09-15 13:56 PDT, Leandro Pereira
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Ryosuke Niwa 2011-09-14 00:40:40 PDT
Instead of each contributor manually figuring out where his/her area of expertise is, we should automatically figure it out from change log entries.
Comment 1 Eric Seidel (no email) 2011-09-14 13:31:05 PDT
This is slightly related to the webkit-patch suggest-reviewers code I suspect. :)  You might look at
http://trac.webkit.org/browser/trunk/Tools/Scripts/webkitpy/tool/commands/queries.py#L47

Even if you don't re-use existing code, it's possible suggest-reviewers will want to be augmented to include your new fanciness.

Also, we've long considered having a bot automatically run suggest-reviewers (some more refined version than current) and auto-cc people on bugs they're likely to wish to review.
Comment 2 Ryosuke Niwa 2011-09-14 13:32:39 PDT
(In reply to comment #1)
> Also, we've long considered having a bot automatically run suggest-reviewers (some more refined version than current) and auto-cc people on bugs they're likely to wish to review.

Yeah, that'll be a really cool feature to build on top of this.
Comment 3 Leandro Pereira 2011-09-15 13:56:59 PDT
Created attachment 107545 [details]
Classifier, proof of concept 1

I won't be able to work on this script until next Monday, so I'm attaching the current proof of concept so it can be improved/reviewed. 

To determine people who actively works on a certain port, it seems decent enough. For more granularity, there's a lot of room for improvements.

One of these improvements would be using a better classifier, probably based on a Naive Bayes classifier or a Perceptron network based on the changelog entry's description, like Ryosuke suggested on IRC. Also, I did not check Eric's code to suggest reviewers, but I'm positive it (or parts of it) could be used as well.
Comment 4 Ryosuke Niwa 2011-11-10 13:31:40 PST
*** Bug 71957 has been marked as a duplicate of this bug. ***