The principal concepts of DMN are:

• standardized external and exchange representations of decision tables, with formal semantics: columns represent influence factors (in) and decision results (out), input cells represent unary predicates, output cells represent (partial) results (named or numeric values), rows represent conjunction of the inputs and separate implications of the results
• ‘business knowledge models’ (BKM): simple data aggregation and conversion expressed in tabular form, rather like Excel formulas
• ‘decision requirements graphs’ (DRG): representation of a decision complex as a directed lattice of data and decision flows from data source inputs, possibly through BKMs, to Decision Tables whose results become data to other BKMs and decision tables.

A ‘decision tree’ can be represented as a DRG with component decision tables for the tree branches, although DMN implementers do not regard this as ideal.

Finally, we come to the overall problem: The requirement for ‘business people’ to be sufficiently careful in their terminology that the formalized models can be a basis for automation of processes and decisions. That is the nominal domain of SBVR.

Edward Barkmeyer