| | |
Summary: Extended Faceted Taxonomies For Web Catalogs
by Yannis Tzitzikas, Nicolas Spyratos, Panos Constantopoulos and Anastasia Analyti
What do you prefer to remember: 1000 individual terms or 3 facets of 10 terms each?
One way of designing a taxonomy for a knowledge domain is by identifying a number
of different aspects, or facets, of the domain and then designing one taxonomy for
each facet. Several studies in information, library and cognitive science have shown
that in almost every knowledge domain we can indeed distinguish a number of facets,
or planes of understanding. A faceted taxonomy is actually a set of taxonomies, called
facets, where a taxonomy is a set of terms structured by a
specialization/generalization relation. Using a faceted taxonomy, the indexing of
objects is done by associating each object with a compound term, i.e. with a
combination of terms coming from different facets. It has been recognized long ago
that a faceted taxonomy has several advantages by comparison to a single hierarchical
taxonomy, such as conceptual clarity, compactness and scalability. For example,
consider two schemes for indexing the objects of a domain, the first using a single
taxonomy consisting of 100 terms, and the second using a faceted taxonomy
consisting of 10 facets each having 10 terms. The first scheme has 100 indexing terms
while the second has 1010
, i.e. 10 billion, compound indexing terms! Although both
schemes have the same storage requirements, i.e. each one requires storing 100 terms,
|
