Visualization HeuristicsNet

RubenS

Hi,

I have 2 questions about the HeuristicsNet Visualization. 

First of all, why is every activity/task duplicated into a "start" and a "complete" activity?

This happens with the HeuristicsNet Array visualization as well as the HeuristicsNet with Annotations visualization. 

I've already tested it with a few different log files I was given.

I already figured out it has something to do with length-one loops when an activity occurs twice (or more) in a row.

I've already read a few papers written by Weijters and Ribeiro about the Heuristic Miner and the given visualizations

didn't have this although there are clearly length-one loops present.

Second, I fail to understand the Semantic Split/Join Points. 

This feature adds a few X objects on the graph. I have noticed someone else already asked this but wasn't given an answer so far.

Combined with my previous question I always get 2 X objects between every duplicated "start" and "complete" activity.

Can anyone clarify this for me?

Thank you for your help in advance!

RubenS

JBuijs

Hi RubenS,

Each activity in a process can go through different states. 2 Examples are starting and completing an activity, this thus results in 2 events. For more information search for 'event lifecycle' in process mining related material and/or the XES documentation.
The reason that the model contains a start and complete is thus because your eventlog contains this. You can filter this out of the event log by executing the 'filter log using simple heuristics' plug in which should be installed by default.

Adding these split and join points, which are in BPMN notation by the way, is done in a 'stupid' way, just adding them before and after each activity. It might help if you translate the model to a Petri net and interpret that one.

Nami-e15

Hi, Thank you for this topic. I want illustrate the results of Genetic miner in the form of heuristic net i.e. a graph that has a node for each activity and each node has input/output components with xor/And labels. What plugin do this illustration ? What plugin convert a heuristic net to a Petri-net? Now I have only two options for illustration. One is heuristic net array visualizer , another is heuristic net visualizer with split/join semantics.

JBuijs

Dear Nami,

Could you please clarify which version of ProM (e.g. ProM 5.2, 6.1 - 6.6, or ProM lite 1.1) you are using? This influences the answer

Nami-e15

I am experiencing ProM in developing side. before I was using Prom 6.4 but it did not contain a Plugin for Genetic Miner. I also use ProM lite but I don't know its version. ProM lite contain Heuristic Miner plugin which show the Heuristic Net with annotations. 

JBuijs

Dear Nami,

The genetic miner was indeed removed from ProM because we could not maintain it anymore and another genetic algorithm, the ETM, was developed, so having 2 genetic algorithms was considered too much.

You will get more heuristics visualizations if you launch the ProM package manager from the development side (if you're using Eclipse then there should be a launch configuration for this in your package).

Hope this helps.

Nami-e15

thanks very much, but I want to show the routing of tokens in the Heuristic Net.
what software do you recommend to draw petri Net and Heuristic Net ?

JBuijs

I just checked but this visualization was removed since it was not that clear.

However, there should be a plug-in that is able to translate a Heuristics net to a Petri net. While viewing the heuristics net, press the 'play'-button on the top right and the converter plug-in should be listed and ready to be executed.

Nami-e15

I checked it in ProM Lite.  the converter you mentioned, was listed. The petri net that this plugin build is very complex, it is better that a heuristic net be shown by input output components (like figure below).Of course, it is my preference.
 Joos, I thank you for greet support.

fmannhardt

I'm reading this thread just now. Recently, I rewrote the Heuristic Miner and created a new visualization for Causal Nets (or Heuristic Nets as they are called by the Heuristic Miner). The new plug-in is in the package DataAwareCNetMiner (sorry for the cryptic name): Interactive Heuristic Miner.

It creates Causal Nets that can be easily converted to Petri Nets, but also be visualized as shown in the Process Mining book. I hope this still helps someone :-)

Please note that this is currently only available in the nightly builds.