lib/twitter_ebooks/model.rb in twitter_ebooks-2.0.4 vs lib/twitter_ebooks/model.rb in twitter_ebooks-2.0.5

@@ -38,19 +38,12 @@
       sentences = NLP.sentences(text)
 
       log "Tokenizing #{sentences.length} sentences"
       @sentences = sentences.map { |sent| NLP.tokenize(sent) }
 
-      log "Building markov model"
-      @markov = MarkovModel.build(@sentences)
-
       log "Ranking keywords"
-      require 'benchmark'
-      puts Benchmark.measure {
-        @keywords = NLP.keywords(@sentences)
-        p @keywords.top(100)
-      }
+      @keywords = NLP.keywords(@sentences)
 
       self
     end
 
     def save(path)
@@ -73,14 +66,14 @@
 
       NLP.htmlentities.decode tweet
     end
 
     def markov_statement(limit=140, markov=nil)
-      markov ||= @markov
+      markov ||= MarkovModel.build(@sentences)
       tweet = ""
 
-      while (tweet = markov.generate) do
+      while (tweet = markov.generate(@sentences)) do
         next if tweet.length > limit
         next if NLP.unmatched_enclosers?(tweet)
         break if tweet.length > limit*0.4 || rand > 0.8
       end
 
@@ -110,12 +103,9 @@
     # Generates a response by looking for related sentences
     # in the corpus and building a smaller markov model from these
     def markov_response(input, limit=140)
       # First try 
       relevant, slightly_relevant = relevant_sentences(input)
-
-      p relevant
-      p slightly_relevant.length
 
       if relevant.length >= 3
         markov = MarkovModel.new.consume(relevant)
         markov_statement(limit, markov)
       elsif slightly_relevant.length > 5