On second thought, we can't use a cache system

Simply because the corpuses are too darn big to keep around

lib/twitter_ebooks/model.rb

@@ -4,33 +4,14 @@
 require 'json'
 require 'set'
 require 'digest/md5'
-require 'fileutils'
 require 'csv'
 
 module Ebooks
   class Model
     attr_accessor :hash, :tokens, :sentences, :mentions, :keywords
 
-    # Consume a corpus file to create a model
-    # @param corpus_path Path to a json, text or csv file to consume
-    # @param cache Optional path to a directory to store cached models
-    def self.consume(corpus_path, cache: nil)
-      if cache
-        FileUtils::mkdir_p cache
-
-        cache_path = File.join(cache, Digest::MD5.file(corpus_path).to_s)
-        if File.exists?(cache_path)
-          log "Reading model from cache at #{cache_path}"
-          return Model.load(cache_path)
-        end
-      end
-
-      model = Model.new.consume(corpus_path)
-
-      if cache
-        log "Caching model at #{cache_path}"
-        model.save(cache_path)
-      end
+    def self.consume(txtpath)
+      Model.new.consume(txtpath)
     end
 
     def self.consume_all(paths)