Adding Constraint Tables to the DMN Standard: Preliminary Results

Abstract

The DMN standard allows users to build declarative models of their decision knowledge. The standard aims at being simple enough to allow business users to construct these models themselves, without help from IT staff. To this end, it combines simple decision tables with a clear visual notation. However, for real-life applications, DMN sometimes proves too restrictive. In this paper, we develop an extension to DMN's decision table notation, which allows more knowledge to be expressed, while retaining the simplicity of DMN. We demonstrate our new notation on a real-life case study on product design.

Tackling the DM Challenges with cDMN: A Tight Integration of DMN and Constraint Reasoning

“…It has been recognized that even though DMN has many advantages, it is somewhat limited in expressiveness (Calvanese et al 2019;Deryck et al 2019). This holds especially for decision tables with S-FEEL, the fragment of FEEL that is considered most readable.…”

“…The semantics of simple constraint tables (without quantification and functions) is also a conjunction of implications, as we described in (Deryck et al 2019). The semantics of constraint tables and decision tables differ in the interpretation of incomplete tables: when no rows are applicable in constraint tables, its outputs can take any arbitrary value instead of being forced to null (or some default value).…”

“…In Deryck et al (2019), we presented a preliminary framework for constraint modelling in DMN. In the current paper, we extend this by adding quantification, types, functions, relations, data tables, optimization and by evaluating the resulting cDMN formalism on the DMN challenges.…”

Rules and Reasoning

“…It has been recognized that even though DMN has many advantages, it is somewhat limited in expressivity [1,3]. This holds especially for decision tables with S-FEEL, the fragment of FEEL that is considered most readable.…”

“…In [3], the semantics of simple constraint tables (without quantification and functions) is introduced, which is also a conjunction of implications. The semantics of constraint tables and decision tables differ in the interpretation of incomplete tables: when no rows are applicable in decision tables, the output is forced to null (i.e., the implicit default value is null ), while the output in constraint tables can take any value.…”