How to convert PetriNet to PetriNet？

a_lovely_boy

Hello, I want to realize the conversion between org.processmining.models.graphbased.directed.petrinet.Petrinet and org.processmining.framework.models.petrinet.PetriNet, what plug-in can I use, which package is this plug-in in?

hverbeek

Hi,

As far as I know, there are no conversion plugins yet between these two classes. The former class is used in ProM 6, the latter in ProM 5.2 (and earlier) for Petri nets.

Which package are you using in ProM 6 that it requires the Petri net implementation from ProM 5.2?

Kind regards,

Eric.

a_lovely_boy

Hi, Not PetriNet that I generated with ProM 6.This is because I have read a paper, and I have read the code provided by the author. PetriNet is used in it, but AcceptingPetriNet is output through operation at last. So I would like to ask whether there is such a plug-in.Private static void writeModelToFile(PetriNet Model, String fileName) { File file = new File(fileName); try { file.createNewFile(); FileOutputStream outputStream = new FileOutputStream(file); Object[] objects = new Object[] { model }; ProvidedObject object = new ProvidedObject("temp", objects); PnmlExport exportPlugin = new PnmlExport(); exportPlugin.export(object, outputStream); outputStream.close(); } catch (IOException e) { // TODO Auto-generated catch block e.printStackTrace(); } Kind regards, wk.

hverbeek

HI,

Can you provide some details on the paper you mention?

Although the PetriNet class is from ProM 5.2, the AcceptingPetriNet class is from ProM 6. Possibly, the author of the paper has created such a plugin, but then it would help if I knew where to start looking.

Kind regards,

Eric.

a_lovely_boy

Hi, In this paper, the log is fully loaded into memory and PetriNet is returned by mining algorithm.Code details:public class ILPMiner { public PetriNet mine(LogReader logReader) { if (logReader != null) { ParikhLanguageRegionMiner miningPlugin = new ParikhLanguageRegionMiner(); PetriNetResult result = (PetriNetResult) miningPlugin.mine(logReader); return result.getPetriNet(); } else { System.err.println("Log reader is null.空"); return null; } } } The imported packages are: import org.processmining.framework.log.LogReader; import org.processmining.framework.models.petrinet.PetriNet; import org.processmining.mining.logabstraction.LogRelations; import org.processmining.mining.petrinetmining.PetriNetResult; ----------------------- PnmlExport exportPlugin = new PnmlExport(); exportPlugin.export(object, outputStream);The imported packages are: org.processmining.exporting.petrinet.PnmlExport Version:1.0Author:Peter van den Brand I don't know why the word newline I input is invalid. I'm really sorry for causing trouble to your reading. Kind regards, wk

hverbeek

Hi,

The ParikhLanguageRegionMiner is from ProM 5.2. Also, Peter van den Brand has not been working on process mining for some time now. I guess this paper is quite old, and not really up-to-date anymore.

It would help if you could give me a link to this paper, or its title at least.

Kind regards,

Eric.

a_lovely_boy

Hi， The title of the paper is Decomposed and Parallel Process Discovery: A Framework and Application.Link is https://doi.org/10.1016/j.future.2019.03.048 Kind regards, wk.

hverbeek

Hi,

It is unclear for me how you got from this paper to the ILP miner as shown earlier. I scanned the paper, but could not find any reference to the actual implementation of the proposed approach. The authors refer to ProM, but are not clear to which version of ProM they refer.

Perhaps it would be good idea to contact the authors on this.

Kind regards,

Eric.

a_lovely_boy

Hi， Thank you for your patience.There is also a question about log decomposition. What are the meanings of incoming edge and outgoing edge in Activity Clustering, and whether these edges have an impact on log decomposition? Kind regards, wk.

hverbeek

Hi,

Please contact the authors on this as well. This is related to the approach as proposed in the paper, not to ProM.

Kind regards,

Eric.

a_lovely_boy

Hi, Sorry, I didn't make myself clear.In the divide-and-conquer package, there is an activity cluster array. Different clusters and activities are linked by edges. What are these edges?Do these edges affect the projection of activities by the plug-in Split Event Log plug-in? Kind regards, Wk.

hverbeek

Hi,

Ok, thanks. Now I understand.

An edge from a cluster to an activity indicates that in that cluster that activity has incoming edges. An edge from an activity to a cluster indicates that in that cluster the activity has outgoing edges. Like in the example below, Analyze Defect+start has incoming edges in Cluster 0, but outgoing edges in Cluster 1. As such, this activity brings you from Cluster 0 to Cluster 1.

Kind regards,

Eric.

a_lovely_boy

Hi, Thank you for your patient answer. I have another question.In Cluster2, Test Repair+ Complete and Test Repair+start have both incoming edge and outgoing edge. Does this mean that Test Repair+complete and Test Repair+ Start have self-loop?If you use the Split Event Log plug-in, do these edges affect the results? Kind regards, Wk.

hverbeek

Hi,

No, this means that you stay in that cluster when these transitions are executed.

Kind regards,

Eric.

a_lovely_boy

Hi, When I use the divide and conquer plugin, I have some questions to ask you.The default value of target cluster size is always the maximum value when dividing log activity sets.If I change the target cluster size, the number of clusters will change. How does this change?Is there any reference for me to learn more about the principles behind these operations? Kind regards, Wk.

a_lovely_boy

Hi,
There's one more thing I'd like to ask you.I read "Divide and Conquer: A Tool Framework for Supporting decomposed Discovery in Process Mining Verbeek, H.M.W.;Munoz Gama, J.; After the article "Van der Aalst, W.M.P.", I also wanted to use Artificial Data in my experiment, but I could not find these data sets. For example, I opened the article cited by DMKD 2006, which only contained methods to generate these data sets, but no specific data sets.Could you please share DMKD 2006, IS 2014 and BPM 2013 with me?
Kind regards,
Wk.

hverbeek

Hi,

If a cluster has reached its limit (the cluster size), then no nodes will be added (where nodes with strong causalities are added first) to that cluster anymore. Instead, a new cluster will appear.

The IS 2014 data set can be found on https://data.4tu.nl/articles/dataset/Single-Entry_Single-Exit_Decomposed_Conformance_Checking_IS_2014_/12713741, the BPM 2013 data set on https://data.4tu.nl/articles/dataset/_Conformance_Checking_in_the_Large_BPM_2013_/12692969. The DMKD 2006 data set can be found on https://svn.win.tue.nl/trac/prom/browser/DataSets/aXfYnZ-logs.

Kind regards,

Eric.

a_lovely_boy

Hi，
Thank you for your patient answer, which made me have a better understanding of this plug-in.I saw "Decomposed Process Mining: The ILP Case H.M.W. Verbeek and W.M.P. van der Aalst" and "Divide and Conquer:a tool framework for supporting decomposed discovery in process mining Verbeek, H.M.W.; Munoz Gama, J.; Van der Aalst, W.M.P. ". Both of these articles mentioned the division of activity clusters, I would like to ask whether the technology in these two articles is maximal decomposition?
Kind regards, 
Wk.

hverbeek

Hi,

Yes, unless stated otherwise, we use the maximal decomposition.

Kind regards,

Eric.

a_lovely_boy

Hi,

Recently I found some code about Alpha miner, these methods take LogReader as input and output a Petri net, I wonder if inductive miner also has methods that take LogReader as input? Kind regards, WK.