class Corpus

Public Class Methods

new(glob, klass) click to toggle source
# File lib/ruby_nlp/corpus.rb, line 4
def initialize(glob, klass)
  @glob = glob
  @klass = klass
end

Public Instance Methods

bigrams() click to toggle source
# File lib/ruby_nlp/corpus.rb, line 31
def bigrams
  ngrams(2)
end
files() click to toggle source
# File lib/ruby_nlp/corpus.rb, line 9
def files
  @files ||= Dir[@glob].map do |file|
    @klass.new(file)
  end
end
ngrams(n) click to toggle source
# File lib/ruby_nlp/corpus.rb, line 21
def ngrams(n)
  sentences.map do |sentence|
    Ngram.new(sentence).ngrams(n)
  end.flatten(1)
end
sentences() click to toggle source
# File lib/ruby_nlp/corpus.rb, line 15
def sentences
  files.map do |file|
    file.sentences
  end.flatten
end
trigrams() click to toggle source
# File lib/ruby_nlp/corpus.rb, line 35
def trigrams
  ngrams(3)
end
unigrams() click to toggle source
# File lib/ruby_nlp/corpus.rb, line 27
def unigrams
  ngrams(1)
end