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.

Extract sublogs from log

Hi everyone, 

I am working on a massive log, the problem that I encounter is by replaying a log on the Petri net, the plugin takes a long time without producing any results.
and to solve this, I decided to divide the log into small sub-logs without filtering,
Is there a plugin in ProM that can do that? or is there another solution? 

Thank you,
H

Comments

  • Hi,

    How do you want to divide the log into sub-logs without filtering? Do you aim to partition the traces from the log over the sub-logs? If so, you might be bale to use the "Partition Log on Activity Sets" plug-in from the LogSkeleton package. You do need to run the ProM NightlyBuild for this, as this plug-in is not yet in any released version of ProM.

    This plug-in partitions the traces in a log on similar activity sets, and returns an array of (sub-)logs. In every sub-log, the activity set (the set of activities that occur in a trace) is the same for every trace.

    Otherwise, if your aim is just to speed up the replay, you might be able to use the "Replay using Decomposition" plug-in. This plug-in splits every trace into sub-traces, and related sub-traces end up in the same sub-log, which will be replayed on the related sub-model. This decomposition preservers perfect fitness, but may provide a too-high fitness score for non-fitting traces (but less than 1.0).

    A third alternative is to use a plug-in that can filter on trace attribute values. But this would require you to run this plug-in for every sub-log you want to have.

    Kind regards,
    Eric.

  • hfkhammash
    edited August 2019
    Hi Eric, 

    Thank you for suggesting these three alternatives,

    What I meant by dividing the log into sub-logs without filtering was to use the log as it is, complete without removing any trace or event, and just to speed up the alignment computations. For doing that I think the second solution is the best. But I am fascinated by the first one; surely I will try it B)

    Thank you again :)
    kind regards
    H.  
Sign In or Register to comment.