WLD (word1$, word2$ [,dummy] ) or

WLD (word1$, word2$, w1, w2, w3 [,dummy] )

Location Ähnlichkeiten

This function calculates the weighted Levenstein phonetic distance between two strings: the smaller the result, the more that the two strings are phonetically similar.

If two strings are found to be identical, then 0 is returned, otherwise a positive integer is returned.

The value of the dummy parameter does not actually matter - if it is present then the function will not distinguish between upper and lower case characters.

The three additional parameters of the second syntax allow you to alter the importance of three possible factors used to calculate the difference between the strings - each parameter should have a positive value:

  • w1: wrong letters
  • w2: strings too short
  • W3: strings too long


100 a$="Sinclair QL": b$="IBM PC": CLS
110 PRINT a$;" <-> ";b$
120 PRINT\WLD(a$,b$), WLD(a$,b$,0)
130 PRINT WLD(a$,b$,1,1,1), WLD(a$,b$,1,1,1,0)
140 PRINT WLD(a$,b$,0,0,0)
150 PRINT WLD(a$,b$,1,2,3), WLD(a$,b$,3,2,1)