ledes.rb

require 'rubygems'
require 'stemmer'
require 'classifier'

class LedeClassifier
  
  attr :classifier

  def initialize(sections, n)
    @classifier = Classifier::Bayes.new(*sections)
    sections.each{|s| train_with(training_wheels(s, n), s)}
  end

  def train_with(wheels, s)
    s = s.downcase
    wheels.each {|lede| @classifier.send("train_#{s}", lede)}
  end

  def training_wheels(section, n)
    Page.find_by_name(section).posts.find(:all, :order => "id asc", :limit => n).map(&:teaser)
  end
end

# tests
sections = ["News", "Viewpoints", "Voices", "Sports"]
@@classifier = LedeClassifier.new(sections, 3000).classifier
@@output = ""

def test_section(name, n)
  @@output << "-" * 80 + "\n" + name.upcase + "\n"
  
  teasers = Page.find_by_name(name).posts.find(:all, :order => "id desc", :limit => n).map(&:teaser)

  teasers.each do |t|
    test_lede_classification(t)
  end 
end

def test_lede_classification(teaser)
  @@output << @@classifier.classify(teaser) + "\n"
end

sections.each{|s| test_section(s, 10)}

puts @@output