Imperfect information : Imprecision - Uncertainty Philippe

We can do everything. T What IS can do toady is highly limited. What we cannot do is Useless. R Useless confused with Cannot Be Done. So... why to worry about Uncertainty? U No future without using Uncertainty. T The Information Systems Radical. H The Uncertainty Zealot. Summary: This report surveys various forms of imperfect data, be it imprecision or uncertainty. To that end, a structured thesaurus is proposed. The models that have been proposed to represent imprecision and uncertainty are briefly presented, and their meanings are discussed. Imperfection, be it imprecision or uncertainty, pervades real world scenarios and must be incorporated into every information system that attempts to provide a complete and accurate model of the real world. But yet, this is hardly achieved by today's information system products. A major reason might be found in the difficulty to understand the various aspects of imprecision and uncertainty. Is there imprecision and uncertainty in the real world? This is an open question. Whatever the answer, it must be recognized that our picture of the world, that corresponds to the only information we can cope with, never reaches perfection. Data as available for an information system are always somehow imperfect. Until recently, almost all aspects of imperfect data were modeled by probability theory but in the last 20 years, many new models have been developed to represent imperfect data. The large number of models reflects the recent acknowledgment that there exist many aspects of imperfection and that probability theory, as good as it is, is not the unique normative model that can cope with all of them.