SAPERION Blog

As posted sometimes more the BPM fanclub became more and more aware that the predominant part of the staff work has no predictable chains of tasks. At the end of a case, surely,  it looks like a process. But looking to the history fo all of the finished process instances you will get the image as the result of a processing mining of the pioneer Wil van der Aalst shown below.

Process Mining with Edge Filtering (Source: Van der Aalst)

Process Mining is a procedure which is crowling through divers data sources to find process structures by hindsight. This method should help to reduce effort in discovering process structures during workshops, interview or the newer techniques of social = collaborative tools like ARISaglin, IBM BPM Blueworks or Signavio Process Editor.

In this field of unpredictable processes a responsible knowledge worker mainly performes the case. In the area of healthcare, welfare and employment agencies there is more and more spoken about the case manager and with this the case management. The person in charge is mostly working in the context of a structured and unstructured content describing the history and the progress of the case. Additionally he has an list of tasks which have to be carried out as known with the state of the case. If needed he takes a new document from a template repository to write a letter for requesting or informing the parties of that case. With such a document may be a little process will be started, e.g. requesting help from a collegue, waiting for approval or the response. User of Enterprise Content Management Systems may be familiar with such scenarios, which means they are now at the bright sight of the new BPM world, the Case Management.

I have found a beautiful example of a process definition, which we have implemented some years ago. The following graph points up that the variations of coming through a process may hugh:

Process definition with many possible run through variants

The users did have the requirement to nearly forward from any activity to any other activity and with reaptings as well. At least one can model this with PMN 2.0 in a more simple way. A subprocess can consist of acitivities without sequence flows in between them. This means that the users will get these activities into their worklist and can decide which to do depending on the knowlegde about the case. And if the actvitiy is from type looping it can be repeated as well.

Max J. Pucher, a pionieer of the adoptive processe, hast posted an interesting article last week. After years the Gartners and of Fottesters of this world have ignored his patented  solution of User-Trained Agents (UTA), now they are starting to see the basics as part of BPM as well. Agents in his system (Papyrus) are listening what users do. On this history data the user will get suggestions of what to do. Selecting one its relevance is incremented and the others decremented. So the system is adapting itself in respect to the users behavior.