The potential risk of convert causal matrix to petri net?

boboane

Hi, 
I saw the definition of causal matrix in genetic mining. But it's not like petri net or other modelling language, it doesn't talk about the properties like soundness or valid sequences, etc. Does it means it can be converted to petri net safely without mistakes? If not, is it highly similar to causal net or any other modelling languages, I can refer to? I hope to understand all potential problems before this conversion (causal matrix to petri net).

JBuijs

Hi Boboane,

Good question, but you're not quite right I'm afraid.

Most process discovery algorithms work in 2 phases. In the first phase they inspect the event log and build an intermediary data structure. This can be a causal matrix (how many times is a followed by b), directly follows graph, etc. etc. In the second phase this intermediary representation is transtlated to a process model (Petri net, Heuristics net, Causal net, process tree, etc.). So a causal net is simply a registry of what behavior has been seen. Translating this to a process model is actually the key of a discovery algorithm.

I hope my explanation is clear, otherwise please ask.

boboane

JBuijs said:

Hi Boboane,

Good question, but you're not quite right I'm afraid.

Most process discovery algorithms work in 2 phases. In the first phase they inspect the event log and build an intermediary data structure. This can be a causal matrix (how many times is a followed by b), directly follows graph, etc. etc. In the second phase this intermediary representation is transtlated to a process model (Petri net, Heuristics net, Causal net, process tree, etc.). So a causal net is simply a registry of what behavior has been seen. Translating this to a process model is actually the key of a discovery algorithm.

I hope my explanation is clear, otherwise please ask.

Thanks very much for your help JBuijs.