chomchom 0.4.0 → 0.5.0

data/chomchom.gemspec

@@ -25,5 +25,5 @@ Gem::Specification.new do |s|
   s.add_dependency(%q<ruby-readability>)
   s.add_dependency(%q<htmlentities>)
   s.add_dependency(%q<json>)
-  s.add_dependency(%q<grammar_police>)
+  s.add_dependency(%q<grammar_cop>)
 end

data/lib/chomchom/scorer.rb

@@ -1,5 +1,5 @@
 #coding: utf-8
-require 'grammar_police'
+require 'grammar_cop'
 require 'lingua/stemmer' #https://github.com/aurelian/ruby-stemmer
 
 module Chomchom
@@ -45,7 +45,7 @@ module Chomchom
       ss = summary.downcase.split_sentences.each { |s|
         #take a flat 30% for every copied sentence
         copy_taxes.push(0.3) if text.downcase.index(s.gsub(/[^\p{Word}]+$/,'').gsub(/^[^\p{Word}]+/,''))
-        proper_sentences += 1 if GrammarPolice::Sentence.count_linkages(s, dictionary, parse_options) > 0
+        proper_sentences += 1 if GrammarCop::Sentence.count_linkages(s, dictionary, parse_options) > 0
         coverages.push(find_coverages(s, tss, stem_topics)) if s.match(/\p{Word}+/)
       }
       
@@ -86,17 +86,17 @@ module Chomchom
     
     private
     #progressive length tax
-    #max = .025 + .05 + .1 = .175 (17.5%)
+    #max = .02 + .04 + .06 = .12 (12.0%)
     #no punishment for short summary b/c itself won't be able to cover as much
     def length_tax(summary_size)
       if summary_size <= 100
         0
       elsif summary_size <= 200
-        (summary_size-100)*0.00025
+        (summary_size-100)*0.0002
       elsif summary_size <= 300
-        0.025 + (summary_size-200)*0.0005
+        0.02 + (summary_size-200)*0.0004
       else
-        0.025 + 0.05 + (summary_size-300)*0.001
+        0.02 + 0.04 + (summary_size-300)*0.0006
       end
     end
     

data/lib/chomchom/version.rb

@@ -1,3 +1,3 @@
 module Chomchom
-  VERSION = "0.4.0"
+  VERSION = "0.5.0"
 end

data/tests/scoring.rb

@@ -18,10 +18,10 @@ agent = Mechanize.new
 agent.user_agent = "chomchom request client"
 fake_summary = "A look into Al Jazeera's difficulties in breaking into the US market, because they are the terrorist mouthpiece post 9/11. However, there are a handful of small cable companies who are unpatriotic by broadcasting it, and it hasn't been well received for its fresh and informative worldview coverage."
 
-dict = GrammarPolice::Dictionary.create("en")
-opts = GrammarPolice::ParseOptions.parse_options_create
-GrammarPolice::ParseOptions.parse_options_set_linkage_limit(opts, 10)
-GrammarPolice::ParseOptions.parse_options_set_short_length(opts, 5)
+dict = GrammarCop::Dictionary.create("en")
+opts = GrammarCop::ParseOptions.create
+GrammarCop::ParseOptions.set_linkage_limit(opts, 10)
+GrammarCop::ParseOptions.set_short_length(opts, 5)
 urls.each_with_index do |url, i|
   if i==i
     agent.get(url)
@@ -47,6 +47,4 @@ urls.each_with_index do |url, i|
     puts passage = doc.center_of_gravity()
     puts scorer.score(text, passage, topic_words, dict, opts)
   end
-end
-GrammarPolice::ParseOptions.parse_options_delete(opts)
-GrammarPolice::Dictionary.destroy(dict)
+end

metadata

@@ -2,7 +2,7 @@
 name: chomchom
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 0.4.0
+  version: 0.5.0
 platform: ruby
 authors: 
 - Quan Nguyen
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-05-04 00:00:00 Z
+date: 2011-05-08 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: mechanize
@@ -79,7 +79,7 @@ dependencies:
   type: :runtime
   version_requirements: *id006
 - !ruby/object:Gem::Dependency 
-  name: grammar_police
+  name: grammar_cop
   prerelease: false
   requirement: &id007 !ruby/object:Gem::Requirement 
     none: false