ORION: Managing Uncertain (Sensor) Data

An important quality of sensor data is that it is often uncertain or imprecise. This uncertainty can be an inherent aspect of the data (e.g. due to known errors in the measuring device, such as the Gaussian error in GPS readings), or it may be introduced in order to achieve scalability [2, 1], or to ensure a certain level of privacy [4]. Existing database management systems provide virtually no support for handling imprecise or uncertain data. It is possible to store some types of data uncertainty as an extra attribute in a table, however, all the necessary functionality needs to be provided by the user. Furthermore, as discussed below, supporting uncertain data requires the addition of significant new functionality that is not easy to achieve simply through user-defined functionality. In addition to sensor data, uncertainty naturally arises in a large number of applications including scientific data, web data integration, machine learning, and information retrieval. Consequently, the problem of uncertain data management is receiving renewed attention from the database community. The ORIO