A First Step towards an Epistemology of the Subjective Uncertainty: Its Premises and Modeling

The uncertainty may be divides it into two major groups, "objective uncertainty" and "subjective uncertainty". The objective uncertainty has already been extensively explored in works on classic probability, but this is not the case with subjective uncertainty, a field of knowledge where there is not even a well established epistemology. In this work we show a first step towards an epistemology of the subjective uncertainty, its premises, and also a way to model them. If a system fails to model all the premises it will not be able to generically model the subjective uncertainty. This is what happens, for example, with the rules of combination until now used within the Dempster-Shafer Theory, one of the formal models that deals with subjective uncertainty. The Dempster-Shafer Theory provides a method for combining evidence from different sources without prior knowledge of their distributions, however, it has some pitfalls caused by an incomplete modeling of the premises. In this paper we present a method that models the three subjective uncertainty premises (the explicit lack of knowledge, the conflict among the evidence, and the non- uniqueness of the assignment of belief and relative division of the belief among the hypotheses chosen), extending the Dempster-Shafer Theory, correcting its counterintuitive behavior, and allowing its use in a broad range of situations.