Im building an app, I need to compare lots of data, essentially keyword phrases, but I could work with just single keywords if I had too. I need a good fast reliable and fairly cheap way to compare the distance or similarity between key phrases. I've been looking at various concepts such as wordnet , which looks Orishas thouhgh it's only for single keywords, rather complicated with the various different filters and not extremely accurate, it seems to be somewhat useful though not as good as I was expecting. Is there a better API service which you can recommend in the natural language processing sector? Thanks
Lucene is the one-stop solution for what you want, but it does require some setup on your end. If that's overblown, what you should compare would be either Levenshtein distance or something like Metaphone or a stemming algorithm. All have implementations in many languages, and are easily integrated into whatever you're doing.