Aggregation of Subjective Probabilities Based on Logical Information Source Model

Abstract

Uncertainty combination is a type of reasoning for solving problems in the real world being full of uncertainty. It has been studied from a viewpoint of information fusion in the fields of epistemic uncertainty such as Dempster-Shafer theory, possibility theory, and the Certainty Factor model. In the field of probability, however, aggregation of subjective probabilities has been studied in a different aspect of opinion pooling; several individuals or experts each assign probabilities to some events and the assignments are aggregated into a single collective probability. In the context, the probability aggregation should be discussed in viewpoints of agreement and/or fairness even in mathematical approaches. The paper proposes a probability combination from the viewpoint of information fusion using a Logical Information Source Model with causation events, which was proposed for possibilistic aggregation recently.