MAP is less relevant with the new experiments. First, it does not take into account the importance of the error, just whether true positives are found before false positives. Second, we no longer insert multiple errors; thus precision/recall numbers convey more or less the same information.
MAP is less relevant with the new experiments. First, it does not take into account the importance of the error, just whether true positives are found before false positives. Second, we no longer insert multiple errors; thus precision/recall numbers convey more or less the same information.