Jacob did not present Constraint Optimization as we grew used to. Instead, he shared his work on a “Connecting the Dots” approach applied to Mortgage Origination.
He started the talk with a practical use case: Peter applies for a loan but his income is not sufficient so he gets backing from Joe and Dawn allowing him to get the loan. While processing the paperwork, clerk notices that Joe has a business loan with Bill Smith that invalidates the decision. Fortunately Bill has equity on his house that can be put on the table. But in the end, Bill’s son, Tommy, has actually used up part of that equity leading to a small shortfall but still a shortfall.
Exciting real-life adventure! Though I long for a happy ending here…
The fact is that reality can actually be *that* convoluted and Decision Systems should be able to cope with such progressive path to uncovering the facts. If you have applied for a loan, you may have been experiencing similar “day by day” requests from the lending organization.
Jacob assembled together a number of technologies to accommodate for the dot-by-dot decision. It boils down to a state machine that is invoked in a pub-sub architecture. The traditional Decision Management pieces we expect are the rules engine, CEP module and of course a Constraint Solver. Personally I would recommend adding some Analytics in there too for proactive fact discovery — maybe text mining or something like it.
Jacob started and closed the talk with applicability for the Intelligence arm of the Government. Connecting the dots is something we wish we could have done prior to terrorist attempts like underwear bomber Umar Farouk Abdulmutallab.
Learn more about Decision Management and Sparkling Logic’s SMARTS™ Data-Powered Decision Manager
