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.

Black Transitions in Petri net Model

Hello everyone,

I have an event log and I have been trying to process discovery with this log in ProM 6.9. Firstly I used "Alpha Miner" but the process seems like below even if A placed before D and H placed before J activities in the event log. It did not show the connection between them.


Then I used "Mine Petri net with Inductive Miner" and it generated a more sensible process model below. But, at this time, it generated 3 transition elements which are not placed in event logs. When I look at the .pnml file generated to check the difference, there were 3 transitions that look like the transition below. They not filled with "#FFFFFF" color and their activity properties equal to "$invisible". Also, their name information was kind of "tau join", "tau split", etc.

Also, the BPMN format, which I converted in ProM, of the process above seems like below. In there, I believe that there must be a Parallel Gateway after activity A where I drew. Activity A must be connected to B, D, and E activities on this Gateway.

My questions are;
  1. What is the possible reason for the "Alpha Miner" to ignore the connection?
  2. What are these black transitions and Why do they generate?
  3. Must any Parallel Gateway be placed as I drew at last BPMN Diagram? It is about the implementation of the conversation algorithm I used or being more than 2 activities after activity A.
Best regards,
Burakcan
24.png 30.9K

Comments

  • Dear Burakcan,

    1. The Alpha Miner only considers the directly follows relation. If A is never directly followed by D, then A and D will not be connected by a place.
    2. The black transition are so-called silent(or invisible) transitions. They are not related to any event in the log, but are needed for routing purposes (to model the parallelism between F and G and the fact hat this part can be skipped.
    3. No, but I do admit that it makes it clearer. By default, an activity in BPMN acts as an exclusive gateway on its inputs and as a parallel gateway on its outputs. As A has multiple outputs, it acts as a parallel gateway on them, and there is no real need to add an explicit parallel gateway. The parallel gateway before J is required, as it would otherwise be an exclusive gateway.
    Kind regards,
    Eric.
  • Hi Eric, thanks a lot for your response. I check my event logs right after read your response to check that whether A is directly followed by D. All 3 activities (B, D, and E) follow A directly so I can't figure it out.

    On the other hand, I would like to ask something to make it clearer for me. The Parallel Gateway that I asked in question 3 can be added but, at the same time, it can not be added. It is not essential, right? 

    Best regards,
    Burakcan
  • Hi Burakcan,

    Is there also a trace where D is directly followed by A? Otherwise, feel free to send me the log (h.m.w.verbeek@tue.nl), so I can have a look.

    Yes, indeed, it is not essential. nevertheless, adding it makes it more clear as not everybody knows that implicitly every task is a XOR-join and an AND-split. In a similar way, I would also add a XOR-join before H.

    Cheers,
    Eric.
Sign In or Register to comment.