A Model for the Management of Highly Variable Data

The goal of this paper is to present a data model to manage highly variable data of the real world. The management of such data raises new database problems. These problems are the data volume, the great number of states that database operators have to manipulate, as well as the deficiency of versions that leads to non-determinist temporal queries. To this end, a new approach to represent and to manipulate these data is proposed. The representation model consists in describing how a value evolves in the course of time. We associate semantics to this description. Concerning the manipulation, this descriptive representation is converted to an internal polynomial representation. Queries on highly variable data are solved by using numerical algorithm techniques on the polynomial representation. We also present some implementation issues and results of performance evaluations. The paper reports on this ongoing experiment.