File Exchange

## Find nearest-matching string from a set

version 1.2.0.0 (2.43 KB) by Brandon Kuczenski

### Brandon Kuczenski (view profile)

Computes shortest distance between a key string and a set of strings to find the nearest match.

Updated 19 Aug 2013

Uses a modified Vagner-Fischer algorithm to find the Levenshtein distance between each pair of strings. Progressively narrows the threshold to equal the distance of the best match found so far, reducing run time.

Updated to correct the algorithm. (sorry!)

Optional behaviors include an upper-bound threshold distance, detect first or multiple matches of equal distance, case insensitivity.

Examples:

>> [i,d]=strnearest({'first string'},{'string 2','abcdefgh','FURSrtd','firststring'})

i =



d =

2

>> [i,d]=strnearest({'1','first string'},{'string 2','abcdefgh','1st string','FURSrtd','seconaaad string','2'})

i =

 

d =

1 5

>>

Yung-Yeh Chang

### Yung-Yeh Chang (view profile)

Oh...I see

It's actually the Levenshtein distance. The title "nearest-matching string" really confuses me.

Yung-Yeh Chang

### Yung-Yeh Chang (view profile)

This is weird, I tried

[x,d] = strnearest('ABC',{'""','"ABC"'})

but it return
x = [1 2]
d = 3

Brandon Kuczenski

### Brandon Kuczenski (view profile)

apologies- I have uploaded a corrected submission.

@@ -98,13 +98,16 @@
else
bb=list{T};
end
- luma=numel(key); lima=numel(bb);
- dl=dist*ones([luma+1,lima+1]);
- dl(1,:)=0:lima; dl(:,1)=0:luma;
+ eql=char({key;bb}); % equal length
+ luma=numel(keyy);
+ dl=dist*ones([luma+1,luma+1]);
+ dl(1,:)=0:luma; dl(:,1)=0:luma;
%Distance
for i=1:luma
- for j=max([1,i-dist]):min([lima,i+dist])
- kr=krk*(~(key(j)==bb(i)));
+ for j=max([1,i-dist]):min([luma,i+dist])
+ kr=krk*(~(keyy(min([j luma]))==bb(i)));
dl(i+1,j+1)=min([dl(i,j)+kr,dl(i,j+1)+1,dl(i+1,j)+1]);
end
end

lala

### lala (view profile)

Still no working example? :)

Sven

### Sven (view profile)

>> [index,distance] = strnearest({'first string'},{'string 2'})

index =



distance =

0

Perhaps a working example would be useful.