A question about IM

XuanSu

Dear process mining community,

Specifically, my question pertains to two methods for mining Petri Nets from event logs . The first method involves using the "Mine Petri Net with Inductive Miner" plugin directly on the log to obtain a Petri Net. The second method involves first using the "Convert log to directly follows graph" plugin to convert the log into a DFG, and then using the "Mine Petri net with Inductive Miner - directly follows" plugin to convert the DFG into a Petri Net.

     While the results from these two methods are sometimes the same, other times they differ. Initially, I assumed that the two methods were based on the same principle, but experimental results suggest otherwise. Can you please provide me with a detailed explanation of the differences in principles between these two methods?

Thank you very much!!

An example experiment result.doc

hverbeek

Hi,

Basically, these two methods use different implementations with different series of cut finders. As such, the first method is not a combination of both plugins of the second method. The first method uses 5 cut finders) in a certain order(say, 1-2-3-4-5), while the second uses only 4 cut finders, where the last 2 also appear in the other order (say, 1-2-4-3). Why these differences exist, I do not know.

Kind regards,

Eric.

XuanSu

Hi Eric,

Thank you very much for your answer. I have another question, what do you mean by '5'? I only know that the 1，2，3，4 may represent sequence，exclusive，parallel，loop. Can you tell me? Looking forward to your reply.

Thank you very much!!

Best regards,

Xuan

hverbeek

Dear Xuan,

Please see https://svn.win.tue.nl/repos/prom/Packages/InductiveMinerDeprecated/Trunk/src/org/processmining/plugins/InductiveMiner/mining/cuts/IM/CutFinderIM.java for the actual source file: 5 represents CutFinderIMConcurrentWithMinimumSelfDistance (see https://svn.win.tue.nl/repos/prom/Packages/InductiveMinerDeprecated/Trunk/src/org/processmining/plugins/InductiveMiner/mining/cuts/IM/CutFinderIMConcurrentWithMinimumSelfDistance.java).

Kind regards,

Eric.

XuanSu

Dear Eric,

Thanks for your kindly answer. I will carefully study it. ;)

Best regards,

Xuan