How does Pattern Abstraction plug-in represent log transformation?

FraDibbe1

Hi, I'm an Italian student and i'm working, for a project at my university,at Process Mining.

Now i'm working on process discovery using different data sets, the objective is to put on work the main process mining techniques in order to discover the control-flow, organizational and social prospective .

I tried with the data sets on TU/e datacentrum, the call center example of Nitro and the data of prom's tutorial, but i have some difficulties interpreting them (Fuzzy and Heuristic miner in first place).

 

So, i read about the Pattern Abstraction plug-in on "Process Mining in the Large" by J. Bose, and i was wandering if i can use that technique to transform the spaghettiness of real life process (or simulated process with some nested AND/XOR splits like the YAWL model of the book) in different sub-process that are lasagna-like.

I'm using either ProM 5.2 and ProM 6.1; in ProM 6.1 there is the Pattern Abstraction plug-in but is not possible to "Transform Log" in fact if i click on that button a windows came up asking me where i want save the file. But the saved file is without and extension, and if i put the .xes o .mxml extensions and try to import in ProM the import procedure says me that "No process is contained in that file".

So my question is: where could i find a good implementation of this plug-in? 
Because i can find plug-in documentation and a book on this plug-in but i can't believe that is not implemented in ProM for the users yet.

Thank you.

jcbose

Hi,

In case you have access to ProM repository, I would recommend that you try running the plug-in from eclipse using the latest source code. You shouldn't face the issue that you mentioned from the latest source.

There seems to be some issue with respect to the visibility of this plugin in ProM 6.2. I will try to figure it out and resolve it. In the meantime, you can use the plug-in via eclipse as I mentioned earlier.

Best,
JC

FraDibbe1

Edit: i founded this post after i posted mine.

Basically if i click on Transform log and use the extension namefile.xes.gz it works!(on ProM 6.1)

The question now is:

Why if i select an abstraction (in this case with no meaning, for example i chosen [Handle case], how is visible in the attached picture), the log is transformed with those abstractions?

ABS4

ABS5

Handle Case

Inbound Call

When I selected only Handle Case, obviously ?

Those abstractions are included in the transformed log output file, and every trace belonging to an abstraction is extracted and saved in a xes separated file (because they are like sub-process) in a directory containing:

abs4.xes.gz

abs5.xes.gz

[Handle case].xes.gz

[Inbound call].xes.gz


Thank you!