Database Reference
In-Depth Information
(a) Inscribed base octahedron
(b) Level 1 hierarchical tessellation
(c) Level 3 hierarchical tessellation
Figure 6.13
Spherical indexing with HTM (see http://www.sdss.jhu.
edu/htm).
6.5.4 Simple Scanned Access of Multidimensional Datasets
Most of the multidimensional access methods described in preceding sections
are only suitable for low-dimensional datasets of the order of
k
8. For
datasets with higher dimensionality the
curse of dimensionality
sets in. The
term
curse of dimensionality
was coined by Bellman to describe the problem
caused by the exponential increase in volume associated with adding dimen-
sions to a metric space.
10
This has two main implications on an indexing
method. As the number of dimensions of data increases, the index size in-
creases as well. Such indexes are usually only effective for exact-match queries
and range queries where all indexed dimensions are used; if only a few of
≤
