Spidr and Raingrams are back, now with specs
2008 / 11 / 13 — course, crawler, generate, json, library, marshal, ngram, ngrams, obstacle, raingrams, random, rspec, rubygem, rubygems, spec, spider, spidr, text, web
Raingrams is back in action. After sitting on rubyforge for quite some time, I was asked to add some features to the general purpose Ngrams Ruby library. I ended up refactoring the code to handle probability calculations better (only recalculate the Maximum Likelihood Estimation (MLE) when the set of ngrams changes), removed the Unigram model (kinda pointless in a ngrams library), allow a trained model to be dumped to a file using Marshal and added the ability to generate random text from trained models. Raingrams also received a total of 133 new spec tests.
$ sudo gem install raingrams
$ sudo gem install spidr