MVLSC HomeIssue Contents

Ascribing Causality from Observational and Interventional Belief Function Knowledge Modeling
Imen Boukhris, Salem Benferhat and Zied Elouedi

The purpose of this paper is to propose a model that an intelligent agent will use to ascribe causality from his uncertain background knowledge expressed under the belief function framework and a temporal sequence of observations or interventions occurring in his environment. The use of interventions which are external actions that alter the natural behavior of the system allows to distinguish causal relations from spurious correlations. The proposed approach is based on the concepts of acceptance and rejection instead of changes in uncertainty distributions to discriminate between potential causes. More different definitions of acceptance and rejection are introduced allowing the categorization of causes according to their strength. In this model, even accepted, an event can be confirmed or attenuated. Different strengths of facilitation and justification, concepts complementary to the concept of causality, are also discussed in this paper.

Keywords: Causality ascription, uncertainty, belief function theory, causal belief networks, interventions

Full Text (IP)
Supplement (IP)