Some teams got problem I accepted by running a heuristic or a randomized method. Apparently test data didn't cover cases that would break those solutions.
I'm interested in knowing what methods people used on this problem other than the intended bipartite matching solution.