youtube_comments.rb

#!/usr/bin/env ruby

require 'open-uri'
require 'nokogiri'
require 'raingrams'

MODEL_FILE = 'youtube_corpora.ngrams'

model = if File.file?(MODEL_FILE)
          Raingrams::BigramModel.open(MODEL_FILE)
        else
          Raingrams::BigramModel.new
        end

if ARGV[0]
  doc = Nokogiri::HTML(open(ARGV[0]))

  puts "Training from: #{doc.at('#watch-headline-title').inner_text.strip}"

  doc.search('div.comment-text').each do |node|
    model.train_with_text(node.inner_text)
  end

  model.save(MODEL_FILE)
else
  puts model.random_sentence
end