Business Rules Uncertainty Management with Probabilistic Relational Models

Hamza Agli (1) , Philippe Bonnard (2) , Christophe Gonzales (1) , Pierre-Henri Wuillemin (1)

1 DECISION
2 IBM France Lab [Biot]

Object-oriented Business Rules Management Systems (OO-BRMS) are a complex applications platform that provide tools for automating day-today business decisions. To allow more sophisticated and realistic decision-making, these tools must enable Business Rules (BRs) to handle uncertainties in the domain. For this purpose, several approaches have been proposed, but most of them rely on heuristic models that unfortunately have shortcomings and limitations. In this paper we present a solution allowing modern OO-BRMS to effectively integrate probabilistic reasoning for uncertainty management. This solution has a coupling approach with Probabilistic Relational Models (PRMs) and facilitates the inter-operability, hence, the separation between business and probabilistic logic. We apply our approach to an existing BRMS and discuss implications of the knowledge base dynamicity on the probabilistic inference.

Mots clés: Business Rules Management Systems; Uncertainty management; Probabilistic Relational Models; Bayesian Networks

Domaines: Informatique [cs]

Hamza Agli, Philippe Bonnard, Christophe Gonzales, Pierre-Henri Wuillemin. Business Rules Uncertainty Management with Probabilistic Relational Models. RuleML16, Jul 2016, Stony Brook, New York, United States. ⟨10.1007/978-3-319-42019-6_4⟩. ⟨hal-01345421⟩