I noticed some posts here on string matching, which reminded me of an old problem I'd like to solve. Does anyone have a good Levenshtein-like algorithm that is weighted toward Qwerty keyboards?
I want to compare two strings, and and allow for typos. Levenshtein is okay, but I'd prefer to also accept spelling errors based on the physical distance between keys on Qwerty keyboard. In other words, the algorithm should prefer "yelephone" to "zelephone" since the "y" key is located nearer to the "t" key than to the "z" key on most keyboards.
Any help would be great... this feature isn't central to my project, so I don't want to veer off into a rat-hole when I should be doing something more productive.
The cpan contributor Kyle R. Burton has actually implemented this distance function in Perl. He uses a table to calculate the weight. See his docs for the full table.