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.

PM on chess logs, complex processes

Hi, I want to analyse (many) chess games with ProM, but I get very unclear spaghetti models, which also take a very long time to be built. I realise this is because of the complexity of chess, it has a very high concurrency and many different tasks and transitions. Are there any plugins I can use to make the process models more clear, abstract infrequent behaviour or make the discovery process take less time in general?

Comments

  • Hi,

    At what level of details or your event logs? Is an activity a move from one square to another, like "e2e4" (or something like that). If so, then we already have many many possible activities.

    Cheers,
    Eric.

  • My event logs will consist of:
    Game ID, timestamp, player color (black or white) and move notation (Nf3). 
    This is still a work in progress since I haven't found the best way to do this yet, but can easily be altered. Eventually I want to analyse "Winning" patterns. 
  • Psychological issues aside, timestamps are not that relevant, it seems. But that still leaves you with many possible activities and transitions.

    Perhaps it makes sense to start with end games. Less pieces, less activities, and less transitions. This may still be comprehensible, and may give you ideas for possible abstractions.

    Other than that, I would not know where to start.

    Cheers,
    Eric.
  • Thanks Eric! I will start with that. One more thing, I read a paper "Mining Chess Playing as a Complex Process" https://www.researchgate.net/publication/318184836_Mining_Chess_Playing_as_a_Complex_Process, where they used a WoMan framework based on first order logic for complex processes. Do you know if this is also doable with ProM?
  • When do you call something doable :)?

    I've scanned the paper you mention, and the use of Petri nets seems to differ a bit from what regular ProM users are familiar with. Typically, we model activities as transitions, not as tokens. As such, there will not be a lot of support for this kind of modeling. But ProM is extensible, and support could be added (typically by those that need it).

    If this way of modeling helps mining complex patterns like short loops, duplicate tasks, etc., then it may help advance process mining in general (and the use of ProM).

    Cheers,
    Eric.
Sign In or Register to comment.