weighted search
<information science> A search based on frequencies of the search terms
in the documents being searched. Weighted search is often used by search
engines. It produces a numerical score for each possible document. A document's
score depends on the frequency of each search term in that document compared
with the overall frequency of that term in the entire corpus of documents. A
common approach is called tf.idf which stands for term frequency * inverse
document frequency. Term frequency means "the more often a term occurs in a
document, the more important it is in describing that document."
http://ciir.cs.umass.edu/cmpsci646/ir4/tsld034.htm Inverse document frequency
means the more documents a term appears in, the less important the term is.
A simple weighted search is just a list of search terms, for example: car
automobile
Weighted search is often contrasted with boolean search. It is possible to have
a search that syntactically is a boolean search but which also does a weighted
search.
See also query expansion.
For a detailed technical discussion see Chapter 5, "Search Strategies", in the
reference below.
["Information Retrieval", C. J. van Rijsbergen,].
(1999-08-28)
Nearby terms:
weeds « weenie « Weenix « weighted search »
well-behaved » well-connected » well-known port
|
