Chapter 10 INDEXING UNCERTAIN DATA

As the volume of uncertain data increases, the cost of evaluating queries over this data will also increase. In order to scale uncertain databases to large data volumes, efficient query processing methods are needed. One of the key techniques for efficient query evaluation is indexing. Due to the nature of uncertain data and queries over this data, existing indexing solutions for precise data are often not directly portable to uncertain data. Even in situations where existing methods can be applied, it is often possible to build more effective indexes for uncertain data. In this Chapter we discuss some of the recent ideas for indexing uncertain data in support of range, nearest-neighbor, and join queries. These indexes build on standard well-known indexes such as R-trees and/or signature trees. In some cases this involves augmenting the standard indexes with extra information. Sometimes more robust clustering criteria are required to make such indexes efficient.