Class: Corpus

Inherits:
Object
  • Object
show all
Defined in:
lib/ruby_nlp/corpus.rb

Instance Method Summary collapse

Constructor Details

#initialize(glob, klass) ⇒ Corpus 



4
5
6
7
 
# File 'lib/ruby_nlp/corpus.rb', line 4

def initialize(glob, klass)
  @glob = glob
  @klass = klass
end

Instance Method Details

#bigrams ⇒ Object 



31
32
33
 
# File 'lib/ruby_nlp/corpus.rb', line 31

def bigrams
  ngrams(2)
end

#files ⇒ Object 



9
10
11
12
13
 
# File 'lib/ruby_nlp/corpus.rb', line 9

def files
  @files ||= Dir[@glob].map do |file|
    @klass.new(file)
  end
end

#ngrams(n) ⇒ Object 



21
22
23
24
25
 
# 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 ⇒ Object 



15
16
17
18
19
 
# File 'lib/ruby_nlp/corpus.rb', line 15

def sentences
  files.map do |file|
    file.sentences
  end.flatten
end

#trigrams ⇒ Object 



35
36
37
 
# File 'lib/ruby_nlp/corpus.rb', line 35

def trigrams
  ngrams(3)
end

#unigrams ⇒ Object 



27
28
29
 
# File 'lib/ruby_nlp/corpus.rb', line 27

def unigrams
  ngrams(1)
end