The behavior of Flexible Heuristics Miner in ProM 6.5.1

yaguang

Dear Sirs,

  Recently, I fould a little problem about Flexible Heuristics Miner(FHM) in ProM 6.5.1. I think the FHM does not perform well in some situations. But according to the paper about the Flexible Heuristics Miner, such problem should not appear. So I think maybe the program for HM has a problem.
  For the example log which contains 3 types of traces :<A,C,D,F>,<A,D,C,F>,<A,C,E,D,F>, pretend each trace appears 100 times in the log. According to the paper of the FHM, activity A should have both {C,D} and {C} as outputs (I mean after activity A there should be a OR-Split). But the result from the FHM software from ProM 6 seems to be wrong? Im I correct? I have shown the mining result for the example log in the picture bellow.

JBuijs

Hi Yaguang,

Thank you for your question.

There are indeed some issues with the heuristics miner, which are investigated by Felix Mannhardt. See f.i. http://www.win.tue.nl/promforum/discussion/comment/1821/#Comment_1821

At some point in time we might release a new version of the heuristics miner.

yaguang

Thank you:). Looking forward to the new edition.

fmannhardt

Dear Yaguang,

I finished my work on the Heuristic Miner. It is not released in the ProM package DataAwareCNetMiner (sorry for the cryptic name). You can find it in the current nightly build.

It comes with an interactive plug-in: Interactive Heuristic Miner.

Could you try this tool on the event log and see whether it gives the desired output?

fmannhardt

Here is some documentation for the tool:
https://fmannhardt.de/blog/data-aware-heuristic-miner-dhm