Skip to Main content Skip to Navigation
Journal articles

Inférence incrémentale pour les modèles probabilistes relationnels et application aux systèmes à base de règles orientés objet

Abstract : This article investigates the exploitation of probabilistic rules within object-oriented business rule management systems (OO-BRMS). In order to facilitate the modelling of the probability distributions in these systems, we propose to use probabilistic relational models (PRM), which are object-oriented extensions of Bayesian networks. When OO-BRMS are exploited by users, numerous requests are sent to the PRM and their answers need be computed very quickly. For this purpose, we propose a novel algorithm that exploits the specificities of OO-BRMS: i) first, the probabilities of interest concern only a subset of the PRM’s random variables, which are called targets; and ii) successive requests differ only slightly. Our new algorithm, IJTI, exploits these two specificites in order to optimize computations.
Complete list of metadata

https://hal.archives-ouvertes.fr/hal-01864391
Contributor : Pierre-Henri Wuillemin Connect in order to contact the contributor
Submitted on : Wednesday, August 29, 2018 - 5:51:00 PM
Last modification on : Tuesday, January 4, 2022 - 3:51:02 AM

Identifiers

  • HAL Id : hal-01864391, version 1

Citation

Hamza Agli, Philippe Bonnard, Christophe Gonzales, Pierre-Henri Wuillemin. Inférence incrémentale pour les modèles probabilistes relationnels et application aux systèmes à base de règles orientés objet. Revue des Sciences et Technologies de l'Information - Série RIA : Revue d'Intelligence Artificielle, Lavoisier, 2018, Réseaux bayésiens et modèles probabilistes, 32 (1), pp.111-132. ⟨hal-01864391⟩

Share

Metrics

Record views

85