Tuesday, August 26, 2014

The Missing Models in Process

Business Process Management (BPM) claims to be model driven, but why is BPM ignoring key models? We find that process flows, decisions, cases, data and result visualizations are often modeled well in BPM technologies. These models have served processes and business results well to date, but there is much more for processes to model.The importance of semiotic representations of process components can't be overstated.  I have identified another six areas so far, that are under-served at best and not dealt with to date, at the worst. Enumerated are areas that processes will have to deal with as larger scopes and more complexity is introduced to the process world.


















Collaborations:

There are few accepted representations approaches for collaboration patterns, though I have seen some attempts to represent social interactions captured during social mining and analytics. It may take a while for this area to come up with some common patterns of representation.

Constraints:

As processes become less structured and more emergent in nature, constraints will become more important in the process governance and control arena. I have not seen any great representation of boundary constraints that show a process surrounded with these protection mechanisms.

Contexts:

As processes become more proactive in sensing active patterns and scenarios, there will be a need for representing the contexts that process participate in and are interested in for sensing events and patterns. These contexts will be inspected for advantages, threats and learning experiences.

Goals:

While there are couple of great examples of goal models and goal interactions by a few brave vendors, there is no accepted modeling approach that is being embraced today. In fact few organizations have send the turn to goal driven processes from flow directed processes.

Patterns:

Today organizations seem so be more interested in events to react to than patterns of interest. This will change over time, but there is no accepted way to represent patterns today.  Patterns of interest will need semiotic representations eventually.

Poly-Analytics:

While there are some models for singular analytic approaches (PMML for predictive analytics), there is no known modeling method for interacting analytics that one can find in smart processes that exist today.

Net; Net:

There are five established areas for modeling that the process world does take advantage of today, but there are at least six more that are under served. I'm not sure if anyone in the standards groups are focused on the under served areas, but standards scramble to take advantage of known momentum. I suspect we will see some brave vendors stepping up to the challenges here.