1. Eric

    A pity. Having a 200/200 score was not sufficient to win this contest, as the jury liked BPMN models more.

    One juror stated that “the readability is compromised by the quantity of arcs and by the need to add symbols to each”. To me, this seems a bit unfair, as one could also claim that in BPMN the quantity of nodes and the need to add symbols to each compromizes readability.

  2. Eric

    For the 200/200 score, I needed to manually configure a number of log-specific filters and log-specific splitters. We can also use the log skeletons without this manual configurations, making it a fully automated classifier. This automated classifier would have scored 194/200. Especially case 10 is a problem for the automated classifier, as 4 out of the 6 misclassified traces are from this case.

Leave a Reply