summary |
shortlog | log |
commit |
commitdiff |
tree
first ⋅ prev ⋅ next
Matt McCutchen [Mon, 7 Jul 2008 23:37:31 +0000 (19:37 -0400)]
Second version of the reduction.
Significant changes:
- Supports a per-reviewer relative load limit, which is hard with a tolerance.
- Split each reviewer into multiple nodes in order to charge a cost quadratic in
the number of disliked proposals she has to review. Previously, cost was
simply additive.
- Require three reviews per proposal regardless of experts, but provide an
incentive for a knowledgeable review and an additional expert review.
Also, add import and "show-graph" script for graph visualization.
Matt McCutchen [Mon, 7 Jul 2008 23:30:05 +0000 (19:30 -0400)]
Fix bug in second-pass graph construction when a proposal got more than one expert review.
Matt McCutchen [Sun, 6 Jul 2008 15:32:52 +0000 (11:32 -0400)]
- Add the notes Samir emailed me on 2008-07-06.
- Move the program into a subdirectory to cut down on rebuilding.
- Hide the .git contents from Eclipse.
Matt McCutchen [Mon, 23 Jun 2008 22:03:07 +0000 (18:03 -0400)]
- ./debug and ./run now pass along arguments.
- Add ./make-package .
Matt McCutchen [Mon, 23 Jun 2008 21:55:09 +0000 (17:55 -0400)]
The proposal matcher. It works on a small example.