Decision management is an enterprise design approach that puts decisions in the center of the design of an organization. Instead of focusing on rules to design the behavior of organizations and IT systems, decisions should be used. First an analysis is made on where and how the decisions are made. Business rules are input for decisions.

Managing decision means that all aspects of a decision are managed in such a way that the business can understand and verify how decisions are made in their organization. This aspect is important to the governance and compliance obligations that companies are facing these days.

In decision management, decisions are often a decision hierarchy, which resembles the way a backward chaining mechanism infers its result. Since Blueriq has implemented the backward chaining mechanism, there is a good fit between Blueriq and decisions. For that reason we strongly advise to use the ideas of decision management when designing a solution.

There are several methods and notations available on the market that support the implementation of decision management. In this chapter we describe the two most prominent implementations of decision management, Decision Model and Notation (DMN) and The Decision Model (TDM).

Decision Model and Notation (DMN)

Decision model and Notation (DMN) is a model used to describe how decisions are made within an organization. The model is adopted by the Object Management Group (OMG) as one of their standards.

The model consist of two levels, the decision requirements level and the decision logic level. The decision requirements level is a high level structure of decisions, sub-decisions and inputs. The requirements level also has references to logic elements and knowledge sources. The decision logic level describes how different inputs will lead to a single answer for a question. When examining a Blueriq logic model, you can see it also contains the same parts.

Decision requirements level

A decision requirements graph is used to create a graphical high level model on how a decision is made. An example is shown below.

DRG for the calculation of the riskscore for an insurance claim

The model should be read as: The decision riskscore is calculated:

• according to the logic in the RiskScoreCalculation
• and the results from RiskScore Bonus and RiskScore Penalties as input
• using inputdata from the CustomerDossier to calculate the RiskScore Bonus
• In the source ClaimPolicy the policy to calculate the riskscore is described.

The model consist of four different type of shapes and the connections between them. These different shapes are:

Decision logic level

The decision logic level contains logic elements, like decision tables and expressions. These logic elements make it possible to infer a result for a decision. These logic elements are connected to the requirements level using the business knowledge model. The official description of DMN contains a specification of their view on decision tables and an expression language.

The DMN specification allows implementations to use other types of logic elements than the ones proposed in the specification. Since Blueriq already has its own models to create logic, we will be using these for our models. For that reason the logic elements described in the specification will not be described further in this document.

Further information can be found in the specifications here: Beta specification DMN.pdf 

The Decision Model (TDM)

Text