Class: FeatureGenerator

Inherits:
Object
  • Object
show all
Defined in:
lib/libsvm_preprocessor/feature_generator.rb

Instance Method Summary collapse

Constructor Details

#initialize(options = {}) ⇒ FeatureGenerator

Returns a new instance of FeatureGenerator.



7
8
9
10
11
 
# File 'lib/libsvm_preprocessor/feature_generator.rb', line 7

def initialize(options = {})
  @token_map = TokenMap.new
  @options = options
  @options[:mode] ||= :unigram
end

Instance Method Details

#bigrams(ary) ⇒ Object 



42
43
44
 
# File 'lib/libsvm_preprocessor/feature_generator.rb', line 42

def bigrams(ary)
  ary[0...-1].zip(ary[1..-1])
end

#features(ary_of_terms, testing: false) ⇒ Object 



13
14
15
16
17
18
19
20
21
22
23
 
# File 'lib/libsvm_preprocessor/feature_generator.rb', line 13

def features(ary_of_terms, testing: false)
  if @options[:mode] == :unigram
    @token_map.token_map(unigrams(ary_of_terms), testing: testing)
  elsif @options[:mode] == :bigram
    @token_map.token_map(unigrams(ary_of_terms) +
                         bigrams(ary_of_terms),
                         testing: testing)
  elsif @options[:mode] == :trichar
    @token_map.token_map trichar(ary_of_terms)
  end
end

#hash_of_ngrams ⇒ Object 



3
4
5
 
# File 'lib/libsvm_preprocessor/feature_generator.rb', line 3

def hash_of_ngrams
  @token_map.hash_of_ngrams
end

#trichar(ary_of_terms) ⇒ Object 



25
26
27
28
29
30
31
32
33
34
35
36
 
# File 'lib/libsvm_preprocessor/feature_generator.rb', line 25

def trichar(ary_of_terms)
  string = ary_of_terms.join(" ")
  if string.size < 3
    return [ [string] ]
  end
  string1 = string[0...-2].split(//)
  string2 = string[1...-1].split(//)
  string3 = string[2..-1].split(//)
  string1.zip(string2).zip(string3).map do |x|
    [x.flatten.join]
  end
end

#unigrams(ary_of_term) ⇒ Object 



38
39
40
 
# File 'lib/libsvm_preprocessor/feature_generator.rb', line 38

def unigrams(ary_of_term)
  ary_of_term.map { |term| [term] }
end