statistical pattern matching
(gcide) | Pattern matching \Pat"tern match"ing\, n. [See pattern.]
(Computers)
A technique in automated data analysis, usually performed on
a computer, by which a group of characteristic properties of
an unknown object is compared with the comparable groups of
characteristics of a set of known objects, to discover the
idenity or proper classification of the unknown object.
Note: There are two major types of pattern matching,
statistical pattern matching and {syntactic pattern
matching}. In statistical pattern matching, the
criteria used to recognize identity or class membership
vary, but in general some combination of the
differences in the groups of characteristics of known
and unknown objects are considered to be a measure of
the difference ("distance") between them, and the
closest known object or objects are viewed as
presenting the most likely identity or class for the
unknown object. In syntactic pattern matching, a set of
known patterns, e.g. as in the possible order of parts
of speech in a language, is defined, and the unknown
pattern is compared to find that known pattern or
patterns which matches the unknown exactly. In general,
statistical pattern matching is used where properties
of objects with continuous values are being compared,
and syntactic pattern matching where a complex
arrangement of at least two different objects may be
built by application of a set of rules (a "grammar")
for combining the objects in a specified order.
Examples of the latter are natural and formal
languages.
[PJC] |