Rule system analysis


Each line in the summary will be deployed into the web hrefs to the page. 


The use of centralized business rule systems is commonplace in financial services and health care.  It's a strategy that must integrate IT and CAS architectural aspects. 

Centralization of business rules in a repository ensures business process flows can apply an agent like use of sensors to leverage the 'executable' derivative of the plan to select how to proceed.  The use of audit, monitoring, lifecycles and business editing capabilities allows the PDCA is the Plan Do Check Act (PDCA) cycle developed at Bell Labs by Walter Shewhart.  Each activity that an organization identifies it will 'do' is formally described in a 'plan' including a model of the expected outcomes of the act if it successfully achieves its goals.  If a subsequent 'check' of the results of the action relative to the predicted model is unsatisfactory the initial plan is modified to correct for identified problems and the cycle is re-executed.  The cycle allows learning to become represented in the improved plan.  Following the process institutionalizes the learning. 
to interactively improve the schematic plan

The application of a genetic algorithm to improve the viability of the rule set is offset by the centralized ability to alter key aspects of the rules.  Given the need to specialize the rules by facility, business relationship and environment which would seem to benefit from a GA and the goal of central control and architectural understanding which would seem to be undermined by a GA the real implementations are not surprisingly challenged. 

The products like FICO's Blaze Advisor that implement these complex rule sets are like computer programs written in primitive languages like BASIC without type checking, scope control, or memory management support.   Therefore they become brittle and difficult to manage.