To prevent spam users, you can only post on this forum after registration, which is by invitation. If you want to post on the forum, please send me a mail (h DOT m DOT w DOT verbeek AT tue DOT nl) and I'll send you an invitation in return for an account.

Can't replay a Log on PetriNet

sitesv
edited September 2013 in ProM 6
Good day!
I have a problems with plugin "Replay a Log on PetriNet for  Conformance".
No one event of log don't map to transition of PetriNet.

Please, see the attachment: 
1. File PetriNet1.pnml
2. File PetriNet2.pnml
3. Log.mxml

Replaying a Log on PetriNet1 is successful, but on PetriNet2 a have a trouble.
Body of both networks is identical in structure. But something is still wrong.
Someone can answer me, what i do wrong?

Big Thanx!

Answers

  • Hi Sitesv,

    Can't you define the mapping between the event log event classes and the transitions in the petri net yourself?
    Joos Buijs

    Senior Data Scientist and process mining expert at APG (Dutch pension fund executor).
    Previously Assistant Professor in Process Mining at Eindhoven University of Technology
  • sitesv
    edited September 2013
    Hi, JBuijs!
    I didn't do a mapping manually, it was made automatically and correctly. (Please, see screenshots and earlier added the logs and petriNets).

    Screenshots attached:
    1. The configuration process
    2 .Replaying result

    Big thanx!
  • Hi, sitesv!

    There is nothing wrong here. The problem is that PetriNet2 has no complete firing sequence, i.e., no firing sequence to reach the final marking from the initial marking. Therefore, no alignments can be made. Hence, no conformance diagnostics. After firing "Test repair+start", no more transitions can be fired (the net is in deadlock state).

    If you don't care about the completion of model, i.e., fitness is not penalized because of no completion, you can use an algorithm that does not penalize for improper completion. You can choose this in dialog "Screen Shot 2013-09-27 at 22.35.59.png" that you attached. Here, you can see that the transitions after "Test repair+start" are never fired at all (because they can't). 

    I guess you're using the alpha miner to derive the models from the log. This algorithm does not guarantee that the models it generates can reach proper termination. To my knowledge, only some algorithms have such guarantee, e.g., ILP miner and the tree-model process discovery. I suggest you to use those algorithms so you can do further analysis based on both model/event logs.

    Hope this explanations help,

    Regards,
    Arya
  • Thanx, Arya! Big thanx!

    Can you help me about "tree-model process discovery"? Is it in prom 6.3?
  • Tree model process discovery is available in ProM 6.3 in the following two packages:
    Inductive Miner
    Evolutionary Tree Miner (a genetic algorithm, the 'live Pareto' version is highly recommended)
    Joos Buijs

    Senior Data Scientist and process mining expert at APG (Dutch pension fund executor).
    Previously Assistant Professor in Process Mining at Eindhoven University of Technology
Sign In or Register to comment.