Moving from Adjacency Finder to Ruby's Classifier
So... while my adjacency finder was a fun little exercise, one of the primary benefits I've received from it is an interest in seeing what else is out there. I was chatting with mrkris about applications of combining my adjacency work with a bayesian classifier, training it with bodies of text against its own keywords. For a larger set of data, it would be possible for the classifier to return new keywords for a new body of text, even if that body didn't contain the word in question. For example, if the training text has a decently large mentioning of both iPhone and Apple across multiple articles, then a classifier trained with this data could be fed text only mentioning Apple, and it would offer iPhone as a keyword (useful for tagging, for example).
I discovered most of the Classifier gem contains what is basically a better version of my code, but there is no API for one wanting to treat keywords as categories. Might have to play around there.