stemmer.rb in sastrawi-0.1.1

- old
+ new

@@ -2,20 +2,27 @@
 
 require 'sastrawi/stemmer/context/visitor/visitor_provider'
 
 require 'sastrawi/stemmer/filter/text_normalizer'
 
+##
+# Indonesian Stemmer
+# Nazief & Adriani, CS Stemmer, ECS Stemmer, Improved ECS
+
 module Sastrawi
   module Stemmer
     class Stemmer
       attr_reader :dictionary, :visitor_provider
 
       def initialize(dictionary)
         @dictionary = dictionary
         @visitor_provider = Sastrawi::Stemmer::Context::Visitor::VisitorProvider.new
       end
 
+      ##
+      # Stem a string to its base form
+
       def stem(text)
         normalized_text = Sastrawi::Stemmer::Filter::TextNormalizer.normalize_text(text)
 
         words = normalized_text.split(' ')
         stems = []
@@ -25,10 +32,13 @@
         end
 
         stems.join(' ')
       end
 
+      ##
+      # Stem a word to its base form
+
       def stem_word(word)
         if plural?(word)
           stem_plural_word(word)
         else
           stem_singular_word(word)
@@ -41,10 +51,15 @@
         return matches[1].include?('-') if matches
 
         return word.include?('-')
       end
 
+      ##
+      # Stem a plural word to its base form
+      # Asian J. (2007) "Effective Techniques for Indonesian Text Retrieval"
+      # page 76-77
+
       def stem_plural_word(word)
         first_match = /^(.*)-(.*)$/.match(word)
 
         return word unless first_match
 
@@ -69,9 +84,12 @@
           root_first_word
         else
           word
         end
       end
+
+      ##
+      # Stem a singular word to its base form
 
       def stem_singular_word(word)
         context = Sastrawi::Stemmer::Context::Context.new(word, @dictionary, @visitor_provider)
         context.execute
 