I'm to the point of adding fuzzy matching to the Java LambdaMOO reimplementation I'm working on. For simplicity/performance, I think the best way to do so is by adding some new built in functions for it. I've written up some draft documentation
here.
If there is anyone who might ever use this implementation, I'd love to hear your opinion on the API presented.
Beyond that, if you are maintaining a newer version of the C code, and think that your distribution might benefit from fuzzy matching, I'd like your opinion too. (The code I want to put behind this isn't very deep, and should port reasonably easy to C.)
Specifically, I currently have it specified as returning lists containing a few elements about the match: confidence, match method, matched word, original word, and the underlying IPA representations (if used). But match_object() is going to need to return the object that was matched, and match_verb() is going to need to return the object and verb matched. Lists start to get a little cumbersome. My limited understanding is that the LambdaMOO code bases currently in active development have mostly added in support for a Map object. I think a map object would probably be a much cleaner way of returning this information. But it's somewhat non-traditional and there might be actively developed code bases out there I'm not aware of that don't have a Map object. I'd appreciate any guidance people might have.
Cheers,
Jo