[Freddy]

I was taking a look around for fuzzy logic ideas and came across this.  It's nice and useful in situations where you want to check a piece of text against a list of text items but not necessarily need an exact match.

I'm using it to trigger songs (mp3) at the moment.  So for example say I want to play a song by Iggy Pop.  On my system the mp3s have the artist and then the song name for the filename.  So for example I may want my bot to find Iggy Pop - The Passenger.mp3

It's simple enough to pull the extension (mp3) out of the equation and you are left with just the artist & song name.  But, with a standard string compare you would have to type in exactly the artist - song combination as it is used/stored on the system to get a hit.

But what if you typed in Igy Poop the Passanger, or even spoke the words to your PC but it got a bit garbled by the speech recognition.  You would want that to work too right ?  So use some fuzzy pattern matching and test for similarity between the filename and the user input.

You can set a value whereby the system thinks that if the match is good enough it will proceed and play the song with the best match.  And perhaps also a lower threshold when the system asks 'Did you mean Iggy Pop - The Passenger ?'

Does that make sense ?  Here's some more :

• Levenshtein distance Wikipedia.
• Generic Levenshtein edit distance with C#
• Visual Basic Example
• DLL Example

[Sybershot]

Thanks I be able to use this in the future 

[Freddy]

Glad it made sense