wikipedia-vandalism_detection 0.0.1

data/spec/vandalism_detection/features/removed_character_distribution_spec.rb

@@ -0,0 +1,46 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RemovedCharacterDistribution do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::RemovedCharacterDistribution.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the Kullback-Leibler Divergence of the removed text's character distribution" do
+      old_text = Wikipedia::VandalismDetection::Text.new 'old text [[new inserted text]] given dero 9'
+      new_text = Wikipedia::VandalismDetection::Text.new('old text')
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 1.6609633564650683
+    end
+
+    it "returns missing value if no alphanumeric characters are removed" do
+      old_text = Wikipedia::VandalismDetection::Text.new("old text !* [[?]]")
+      new_text = Wikipedia::VandalismDetection::Text.new("old text")
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq Wikipedia::VandalismDetection::Features::MISSING_VALUE
+    end
+
+    it "returns missing value if no text inserted" do
+      old_text = Wikipedia::VandalismDetection::Text.new("text")
+      new_text = Wikipedia::VandalismDetection::Text.new("deletion text")
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq Wikipedia::VandalismDetection::Features::MISSING_VALUE
+    end
+  end
+end

data/spec/vandalism_detection/features/removed_emoticons_frequency_spec.rb

@@ -0,0 +1,36 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RemovedEmoticonsFrequency do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::RemovedEmoticonsFrequency.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the number of removed emoticon words relative to all removed words count" do
+      # inserted: total 6 words, 2 emoticons
+      old_text = Wikipedia::VandalismDetection::Text.new ":) old. It's :P man:Pio, this is X-D."
+      new_text = Wikipedia::VandalismDetection::Text.new ':) old.'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 2.0 / 6.0
+    end
+
+    it "returns 0.0 on emtpy removed text" do
+      old_text = Wikipedia::VandalismDetection::Text.new "Great old. {{speedy deletion}}"
+      new_text = Wikipedia::VandalismDetection::Text.new 'Great old. {{speedy deletion}} :)'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/removed_markup_frequency_spec.rb

@@ -0,0 +1,36 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RemovedMarkupFrequency do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::RemovedMarkupFrequency.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the number of removed markup words relative to all removed words count" do
+      # inserted: total 6 words, 2 markup
+      old_text = Wikipedia::VandalismDetection::Text.new "[[Great]] old. It's [[Great]] man, this is [[amazing]]."
+      new_text = Wikipedia::VandalismDetection::Text.new '[[Great]] old.'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 2.0 / 6.0
+    end
+
+    it "returns 0.0 on emtpy removed text" do
+      old_text = Wikipedia::VandalismDetection::Text.new "Great old. {{speedy deletion}}"
+      new_text = Wikipedia::VandalismDetection::Text.new 'Great old. {{speedy deletion}} [[heading]]'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/removed_pronoun_frequency_spec.rb

@@ -0,0 +1,36 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RemovedPronounFrequency do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::RemovedPronounFrequency.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::FrequencyBase }
+
+  describe "#calculate" do
+
+    it "returns the number of removed pronouns relative to all removed words count" do
+      # total 12 words, 7 pronouns
+      old_text = Wikipedia::VandalismDetection::Text.new "Your old. I was you if You was Me and we are ourselves us."
+      new_text = Wikipedia::VandalismDetection::Text.new 'Your old.'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 7.0 / 12.0
+    end
+
+    it "returns 0.0 on emtpy removed clean text revisions" do
+      old_text = Wikipedia::VandalismDetection::Text.new "Your old. {{speedy deletion}}"
+      new_text = Wikipedia::VandalismDetection::Text.new 'Your old. My inserted.'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/removed_sex_frequency_spec.rb

@@ -0,0 +1,36 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RemovedSexFrequency do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::RemovedSexFrequency.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::FrequencyBase }
+
+  describe "#calculate" do
+
+    it "returns the number of removed sex words relative to all removed words count" do
+      # inserted: total 10 words, 4 biased
+      old_text = Wikipedia::VandalismDetection::Text.new "Penis old. It's Penis man, this is penis, penis a penis."
+      new_text = Wikipedia::VandalismDetection::Text.new 'Penis old.'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 4.0 / 9.0
+    end
+
+    it "returns 0.0 on emtpy removed clean text" do
+      old_text = Wikipedia::VandalismDetection::Text.new "penis old. {{speedy deletion}}"
+      new_text = Wikipedia::VandalismDetection::Text.new 'penis old. Penis new.'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/removed_size_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RemovedSize do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::RemovedSize.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the size of the edit's new removed inserted text sizes" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new '123 456789'
+      new_revision_text = Wikipedia::VandalismDetection::Text.new '123' # 6 removed
+
+      old_revision = build(:old_revision, text: old_revision_text)
+      new_revision = build(:new_revision, text: new_revision_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 6
+    end
+
+    it "returns 0 if no removed text " do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new '123'
+      new_revision_text = Wikipedia::VandalismDetection::Text.new '123 456789' # 0 removed
+
+      old_revision = build(:old_revision, text: old_revision_text)
+      new_revision = build(:new_revision, text: new_revision_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0
+    end
+  end
+end

data/spec/vandalism_detection/features/removed_vulgarism_frequency_spec.rb

@@ -0,0 +1,36 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RemovedVulgarismFrequency do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::RemovedVulgarismFrequency.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::FrequencyBase }
+
+  describe "#calculate" do
+
+    it "returns the number of removed vulgarism words relative to all removed words count" do
+      # total 10 words, 3 vulgarism
+      old_text = Wikipedia::VandalismDetection::Text.new "Old shit. Fuck you, fu*ck you $lut, you and all the others."
+      new_text = Wikipedia::VandalismDetection::Text.new 'Old shit. New shit.'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 3.0 / 10.0
+    end
+
+    it "returns 0.0 on emtpy removed clean text revisions" do
+      old_text = Wikipedia::VandalismDetection::Text.new "Old shit. {{speedy deletion}}"
+      new_text = Wikipedia::VandalismDetection::Text.new 'Old shit. New shit.'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/removed_words_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RemovedWords do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::RemovedWords.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the number of the edit's removed words" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new 'zero one two three four five six' # 6 removed
+      new_revision_text = Wikipedia::VandalismDetection::Text.new 'zero'
+
+      old_revision = build(:old_revision, text: old_revision_text)
+      new_revision = build(:new_revision, text: new_revision_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 6
+    end
+
+    it "returns 0 if no removed text" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new 'zero' # 0 removed
+      new_revision_text = Wikipedia::VandalismDetection::Text.new 'zero one'
+
+      old_revision = build(:old_revision, text: old_revision_text)
+      new_revision = build(:new_revision, text: new_revision_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0
+    end
+  end
+end

data/spec/vandalism_detection/features/replacement_similarity_spec.rb

@@ -0,0 +1,44 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::ReplacementSimilarity do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::ReplacementSimilarity.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns similarity of the deleted text to inserted in exchange" do
+      old_text = Wikipedia::VandalismDetection::Text.new "this is Mr. Dixon"
+      new_text = Wikipedia::VandalismDetection::Text.new "this is Mr. Dicksonx"
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.8133333333333332
+    end
+
+    it "returns 0 on empty old revisions text" do
+      old_revision = build(:old_revision, text: "")
+      new_revision = build(:new_revision, text: "{{speedy deletion}}")
+
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0
+    end
+
+
+    it "returns 0 on empty new revisions text" do
+      old_revision = build(:old_revision, text: "{{speedy deletion}}")
+      new_revision = build(:new_revision, text: "")
+
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0
+    end
+  end
+end

data/spec/vandalism_detection/features/reverted_spec.rb

@@ -0,0 +1,28 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::Reverted do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::Reverted.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::ContainsBase }
+
+  describe "#calculate" do
+
+    ['rvt', 'rvv', 'revert'].each do |term|
+      it "returns 1 if the edit comment includes '#{term}'" do
+        comment = Wikipedia::VandalismDetection::Text.new "#{term} edited"
+        edit = build :edit, new_revision: build(:new_revision, comment: comment)
+
+        expect(@feature.calculate(edit)).to eq 1
+      end
+    end
+
+    it "returns 0 on emtpy comment" do
+      edit = build :edit, new_revision: build(:new_revision, comment: "")
+
+      expect(@feature.calculate(edit)).to eq 0
+    end
+  end
+end

data/spec/vandalism_detection/features/revisions_character_distribution_spec.rb

@@ -0,0 +1,46 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RevisionsCharacterDistribution do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::RevisionsCharacterDistribution.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the Kullback-Leibler Divergence of the inserted text's character distribution" do
+      old_text = Wikipedia::VandalismDetection::Text.new('old text old text is standing here')
+      new_text = Wikipedia::VandalismDetection::Text.new 'old text [[new inserted text]] given dero 9'
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.6312751553366259
+    end
+
+    it "returns missing value if new revision text is empty" do
+      old_text = Wikipedia::VandalismDetection::Text.new("old text")
+      new_text = Wikipedia::VandalismDetection::Text.new("")
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq Wikipedia::VandalismDetection::Features::MISSING_VALUE
+    end
+
+    it "returns missing value if old revision text is empty" do
+      old_text = Wikipedia::VandalismDetection::Text.new("")
+      new_text = Wikipedia::VandalismDetection::Text.new("new text")
+
+      old_revision = build(:old_revision, text: old_text)
+      new_revision = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq Wikipedia::VandalismDetection::Features::MISSING_VALUE
+    end
+  end
+end

data/spec/vandalism_detection/features/same_editor_spec.rb

@@ -0,0 +1,60 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SameEditor do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::SameEditor.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+    context "both contributors are given" do
+      it "return 1.0 in case of the same previous editor" do
+        old_revision = build(:old_revision, contributor: 'Peter')
+        new_revision = build(:new_revision, contributor: 'Peter')
+        edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+        expect(@feature.calculate(edit)).to eq 1
+      end
+
+      it "returns 0.0 in case of another previous editor" do
+        old_revision = build(:old_revision, contributor: '137.163.16.199')
+        new_revision = build(:new_revision, contributor: 'Peter')
+        edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+        expect(@feature.calculate(edit)).to eq 0
+      end
+    end
+
+    context "previous contributor not given" do
+      it "requests the user from Wikipedia API and returns 1 in case of the same previous editor" do
+        old_revision = build(:old_revision, id: '324557983', contributor: nil) # TOmaxer
+        new_revision = build(:new_revision, id: '329962649', parent_id: '324557983', contributor: 'Tomaxer')
+        edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+        expect(@feature.calculate(edit)).to eq 1
+      end
+
+      it "requests the user from Wikipedia API and returns 0 in case of another previous editor" do
+        old_revision = build(:old_revision, id: '328774110', contributor: nil) # 137.163.16.199
+        new_revision = build(:new_revision, id: '328774035', parent_id: '328774110') # ClueBot
+        edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+        expect(@feature.calculate(edit)).to eq 0
+      end
+
+      it "returns missing if old reivision is not available anymore" do
+        # to get api call, see:
+        # https://en.wikipedia.org/w/api.php?action=query&prop=revisions&rvprop=timestamp&revids=325218985
+        # <rev revid="325218985"/>
+
+        old_revision = build(:old_revision, id: '325218985', contributor: nil)
+        new_revision = build(:new_revision, id: '326980599', parent_id: '325218985')
+        edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+        expect(@feature.calculate(edit)).to eq Wikipedia::VandalismDetection::Features::MISSING_VALUE
+      end
+    end
+  end
+end