Approximate String Distances
Compute the approximate string distance between character vectors. The distance is a generalized Levenshtein (edit) distance, giving the minimal possibly weighted number of insertions, deletions and substitutions needed to transform one string into another.
adist(x, y = NULL, costs = NULL, counts = FALSE, fixed = TRUE, partial = !fixed, ignore.case = FALSE, useBytes = FALSE)
- a character vector. Long vectors are not supported.
- a character vector, or
NULL(default) indicating taking
- a numeric vector or list with names partially matching
insertions, deletions and substitutions giving
the respective costs for computing the Levenshtein distance, or
NULL(default) indicating using unit cost for all three possible transformations.
- a logical indicating whether to optionally return the
transformation counts (numbers of insertions, deletions and
substitutions) as the
"counts"attribute of the return value.
- a logical. If
xelements are used as string literals. Otherwise, they are taken as regular expressions and
partial = TRUEis implied (corresponding to the approximate string distance used by
fixed = FALSE.
- a logical indicating whether the transformed
xelements must exactly match the complete
yelements, or only substrings of these. The latter corresponds to the approximate string distance used by
- a logical. If
TRUE, case is ignored for computing the distances.
- a logical. If
TRUEdistance computations are done byte-by-byte rather than character-by-character.
The (generalized) Levenshtein (or edit) distance between two strings
s and t is the minimal possibly weighted number of
insertions, deletions and substitutions needed to transform s
into t (so that the transformation exactly matches t).
This distance is computed for
partial = FALSE, currently using
a dynamic programming algorithm (see, e.g.,
http://en.wikipedia.org/wiki/Levenshtein_distance) with space
and time complexity $O(mn)$, where $m$ and $n$ are the
lengths of s and t, respectively. Additionally computing
the transformation sequence and counts is $O(\max(m, n))$.
The generalized Levenshtein distance can also be used for approximate
(fuzzy) string matching, in which case one finds the substring of
t with minimal distance to the pattern s (which could be
taken as a regular expression, in which case the principle of using
the leftmost and longest match applies), see, e.g.,
distance is computed for
partial = TRUE using tre by
Ville Laurikari (http://http://laurikari.net/tre/) and
corresponds to the distance used by
agrep. In this
case, the given cost values are coerced to integer.
Note that the costs for insertions and deletions can be different, in which case the distance between s and t can be different from the distance between t and s.
A matrix with the approximate string distances of the elements of
y, with rows and columns corresponding to
TRUE, the transformation counts are returned as the
"counts"attribute of this matrix, as a 3-dimensional array with dimensions corresponding to the elements of
x, the elements of
y, and the type of transformation (insertions, deletions and substitutions), respectively. Additionally, if
partial = FALSE, the transformation sequences are returned as the
"trafos"attribute of the return value, as character strings with elements M, I, D and S indicating a match, insertion, deletion and substitution, respectively. If
partial = TRUE, the offsets (positions of the first and last element) of the matched substrings are returned as the
"offsets"attribute of the return value (with both offsets $-1$ in case of no match).
agrep for approximate string matching (fuzzy matching)
using the generalized Levenshtein distance.
## Cf. http://en.wikipedia.org/wiki/Levenshtein_distance adist("kitten", "sitting") ## To see the transformation counts for the Levenshtein distance: drop(attr(adist("kitten", "sitting", counts = TRUE), "counts")) ## To see the transformation sequences: attr(adist(c("kitten", "sitting"), counts = TRUE), "trafos") ## Cf. the examples for agrep: adist("lasy", "1 lazy 2") ## For a "partial approximate match" (as used for agrep): adist("lasy", "1 lazy 2", partial = TRUE)