Decision-making with business rules or structured arguments: Transforming between representations and transferring the stability property

Abstract

Since around the turn of the century, many organisations use business rules to describe their business processes and decisions. The company Blueriq offers a software platform that helps organisations model and apply these business rules. A contrasting method for decision-making uses ASPIC+, a framework for structured argumentation. Recently, the concepts of stability and relevance were introduced for a smaller instantiation of ASPIC+, referred to here as defeasible-firm (DF) ASPIC+. The Netherlands Police have applied DF ASPIC+ in a form for fraud intake, where stability supports the user experience. The fundamental difference between ASPIC+ and the Blueriq rule engine is that ASPIC+ allows for the attack and defeat of derived conclusions, whereas Blueriq was built to serve unambiguous, final decisions.

In this thesis, we compare the Blueriq platform to ASPIC+. First, we formalise the Blueriq platform into a purely Boolean logic representation. Then, we propose several transformations between this formalised version of Blueriq, ASPIC+ with all of its components (full ASPIC+), and DF ASPIC+. For the transformation from Blueriq to ASPIC+ and the continued transformation to DF ASPIC+, we prove that acceptance behaviour is preserved under the transformations. Part of these new definitions is the last-component principle, which is an addition to the more well-known last-link and weakest-link principles for argument ordering. In the last-component principle, ordinary premises and defeasible rules are equally strong. Using this property, we define the notions of stability and relevance for Blueriq.

Stability and relevance could support the explainability of Blueriq systems. Furthermore, the comparison between Blueriq and ASPIC+ is fruitful for parties using either of the systems. Future work could expand the proposed transformations or implement some of the lessons learned into a decision-making system.