wikipedia-vandalism_detection 0.1.0-java

data/spec/vandalism_detection/features/revisions_character_distribution_spec.rb

@@ -0,0 +1,40 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::RevisionsCharacterDistribution do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns the KL-Divergence of the inserted characters distribution' do
+      old_text = Text.new('old text old text is standing here')
+      new_text = Text.new('old text [[new inserted text]] given dero 9')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 0.6312751553366259
+    end
+
+    it 'returns missing value if new revision text is empty' do
+      old_text = Text.new('old text')
+      new_text = Text.new
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq Features::MISSING_VALUE
+    end
+
+    it 'returns missing value if the old revision text is empty' do
+      old_text = Text.new
+      new_text = Text.new('new text')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq Features::MISSING_VALUE
+    end
+  end
+end

data/spec/vandalism_detection/features/same_editor_spec.rb

@@ -0,0 +1,71 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SameEditor do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    context 'when both contributors are given' do
+      it 'returns 1.0 in case of the same previous editor' do
+        editor = 'Peter'
+        old_rev = build(:old_revision, contributor: editor)
+        new_rev = build(:new_revision, contributor: editor)
+        edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+        expect(subject.calculate(edit)).to eq 1
+      end
+
+      it 'returns 0.0 in case of another previous editor' do
+        old_rev = build(:old_revision, contributor: '137.163.16.199')
+        new_rev = build(:new_revision, contributor: 'Peter')
+        edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+        expect(subject.calculate(edit)).to eq 0
+      end
+    end
+
+    context 'when previous contributor is not present' do
+      context 'in case of the same previous editor' do
+        it 'requests the user from Wikipedia API and returns 1' do
+          # contributor: TOmaxer
+          old_rev = build(:old_revision, id: '324557983', contributor: nil)
+          new_rev = build(
+            :new_revision,
+            id: '329962649',
+            parent_id: '324557983',
+            contributor: 'Tomaxer'
+          )
+
+          edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+          expect(subject.calculate(edit)).to eq 1
+        end
+      end
+
+      context 'in case of another previous editor' do
+        it 'requests the user from Wikipedia API and returns 0' do
+          # 137.163.16.199
+          old_rev = build(:old_revision, id: '328774110', contributor: nil)
+          # ClueBot
+          new_rev = build(:new_revision, id: '328774035', parent_id: '328774110')
+          edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+          expect(subject.calculate(edit)).to eq 0
+        end
+      end
+
+      context 'if old reivision is not available anymore' do
+        it 'returns missing' 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_rev = build(:old_revision, id: '325218985', contributor: nil)
+          new_rev = build(:new_revision, id: '326980599', parent_id: '325218985')
+          edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+          expect(subject.calculate(edit)).to eq Features::MISSING_VALUE
+        end
+      end
+    end
+  end
+end

data/spec/vandalism_detection/features/sex_frequency_spec.rb

@@ -0,0 +1,30 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SexFrequency do
+  it { is_expected.to be_a Features::FrequencyBase }
+
+  describe '#calculate' do
+    it 'returns the number of vulgarism words over all words' do
+      # total 6 words, 3 bad
+      old_text = Text.new('Old whatever.')
+      new_text = Text.new('Old whatever. New sex contents. Penis, dildos, boy.')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 3.0 / 6.0
+    end
+
+    it 'returns 0.0 for an emtpy clean text in the new revision' do
+      old_text = Text.new('Old guy.')
+      new_text = Text.new('Old guy. {{speedy deletion}}')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/sex_impact_spec.rb

@@ -0,0 +1,53 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SexImpact do
+  it { is_expected.to be_a Features::ImpactBase }
+
+  describe '#calculate' do
+    it 'returns the impact of sex words of the new revision text' do
+      # 3 sex words
+      old_text = Text.new('Penis, old text dildo, breast it')
+
+      # 4 sex words
+      new_text = Text.new('Penis, old text dildo, breast anal it')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 3.0 / (3.0 + 4.0)
+    end
+
+    it 'returns 0.5 if both revision text have no terms' do
+      text = Text.new('{{speedy deletion}}')
+
+      old_rev = build(:old_revision, text: text)
+      new_rev = build(:new_revision, text: text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.5
+    end
+
+    it 'returns 0.0 for an emtpy clean text in the old revision' do
+      old_text = Text.new('{{speedy deletion}}')
+      new_text = Text.new('anal')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.0
+    end
+
+    it 'returns 1.0 for an emtpy clean text in the new revision' do
+      old_text = Text.new('anal')
+      new_text = Text.new('{{speedy deletion}}')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 1.0
+    end
+  end
+end

data/spec/vandalism_detection/features/size_increment_spec.rb

@@ -0,0 +1,29 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SizeIncrement do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns a negative increment on more removed texts' do
+      old_rev_text = Text.new('123456789') # length 9
+      new_rev_text = Text.new('123') # length 3
+
+      old_rev = build(:old_revision, text: old_rev_text)
+      new_rev = build(:new_revision, text: new_rev_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 3.0 - 9.0
+    end
+
+    it 'returns a positive increment on more removed texts' do
+      old_rev_text = Text.new('123') # length 3
+      new_rev_text = Text.new('123456789') # length 9
+
+      old_rev = build(:old_revision, text: old_rev_text)
+      new_rev = build(:new_revision, text: new_rev_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 9.0 - 3.0
+    end
+  end
+end

data/spec/vandalism_detection/features/size_ratio_spec.rb

@@ -0,0 +1,48 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SizeRatio do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns the ratio of the revisions text sizes' do
+      old_text = Text.new('123456789') # length 9
+      new_text = Text.new('123') # length 3
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 9.0 / (9.0 + 3.0)
+    end
+
+    it 'returns 1.0 for an emtpy text in the new revision' do
+      old_text = Text.new('sample text')
+      new_text = Text.new
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 1.0
+    end
+
+    it 'returns 0.0 for an emtpy text in the old revisions' do
+      old_text = Text.new
+      new_text = Text.new('sample text')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.0
+    end
+
+    it 'returns 0.5 if both revision texts are empty' do
+      old_rev = build(:old_revision, text: Text.new)
+      new_rev = build(:new_revision, text: Text.new)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.5
+    end
+  end
+end

data/spec/vandalism_detection/features/term_frequency_spec.rb

@@ -0,0 +1,32 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::TermFrequency do
+  it { is_expected.to be_a Features::FrequencyBase }
+
+  describe '#calculate' do
+    it 'returns the average relative frequency of inserted words' do
+      # removed [you, I], added [you, you, you, we]
+      # for [you, we] compute frequency in old_text and average for all words
+      # here: [you] is 6x in new text, [we] is 3x in new text of total 10 words
+      # avg = (6/10 + 3/10)/2
+      old_text = Text.new("we\nwe\nyou\nyou\nyou\nI\nyou\nI\n")
+      new_text = Text.new("we\nwe\nyou\nyou\nI\nyou\n''(you''\nyou\nyou\n[[we]])\n")
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq(((6.0 / 10.0) + (3.0 / 10.0)) / 2.0)
+    end
+
+    it 'returns 0.0 on emtpy clean text revisions' do
+      text = Text.new('{{speedy deletion}}')
+
+      old_rev = build(:old_revision, text: text)
+      new_rev = build(:new_revision, text: text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/time_interval_spec.rb

@@ -0,0 +1,56 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::TimeInterval do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns time interval in days from the old to the new revision' do
+      old_timestamp = '2014-11-27T18:00:00Z'
+      new_timestamp = '2014-11-29T06:00:00Z'
+
+      old_rev = build(:old_revision, timestamp: old_timestamp)
+      new_rev = build(:new_revision, timestamp: new_timestamp)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 1.5
+    end
+
+    it 'requests the time from API if no old revisions timestamp is given' do
+      # to get api call, see:
+      # https://en.wikipedia.org/w/api.php?action=query&prop=revisions&rvprop=timestamp&revids=331655535
+      # => 2001-11-29T03:27:34Z
+      new_timestamp = '2001-11-06T13:16:13Z'
+
+      old_rev = build(:old_revision, id: '331655534', timestamp: nil)
+      new_rev = build(
+        :new_revision,
+        id: '331655535',
+        parent_id: '331655534',
+        timestamp: new_timestamp
+      )
+
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 0.5
+    end
+
+    it 'returns missing if the 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"/>
+      new_timestamp = '2011-11-11T01:00:00Z'
+
+      old_rev = build(:old_revision, id: '325218985', timestamp: nil)
+      new_rev = build(
+        :new_revision,
+        id: '326980599',
+        parent_id: '325218985',
+        timestamp: new_timestamp
+      )
+
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq Features::MISSING_VALUE
+    end
+  end
+end

data/spec/vandalism_detection/features/time_of_day_spec.rb

@@ -0,0 +1,16 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::TimeOfDay do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns the time of day as decimal value (hours)' do
+      old_rev = build(:old_revision)
+      new_rev = build(:new_revision, timestamp: '2012-12-09T05:30:36Z')
+
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 5.6
+    end
+  end
+end

data/spec/vandalism_detection/features/upper_case_ratio_spec.rb

@@ -0,0 +1,30 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::UpperCaseRatio do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns the uppercase to all letters ratio of inserted clean text' do
+      old_text = Text.new('text')
+      # 3 uppercase letters of total 4 inserted letters
+      new_text = Text.new('text [[1A 4B6 8Cd]]')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq((1.0 + 3) / (1.0 + 4))
+    end
+
+    it 'returns 0.0 if no text was inserted' do
+      old_text = Text.new('deletion text')
+      new_text = Text.new('text')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/upper_case_words_ratio_spec.rb

@@ -0,0 +1,33 @@
+# encoding: UTF-8
+
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::UpperCaseWordsRatio do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns the uppercase/all words ratio of the inserted cleaned text' do
+      old_text = Text.new('text')
+      # 2 two capital (not numbers!) words of total 4 inserted.
+      # The template {{23A}} is removed while cleaning.
+      new_text = Text.new('text [[HELLO you]] NICE boyß3 1990 {{23A}}')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq((1.0 + 2) / (1.0 + 4))
+    end
+
+    it 'returns 0.0 if no text inserted' do
+      old_text = Text.new('DELECTION text')
+      new_text = Text.new('text')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/upper_to_lower_case_ratio_spec.rb

@@ -0,0 +1,30 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::UpperToLowerCaseRatio do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns the upper to lower letters ratio of the inserted text' do
+      old_text = Text.new('text')
+      # 3 uppercase letters, 4 lowercase letters
+      new_text = Text.new('text [[1aA 4B6 8Cd ef]]')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq((1.0 + 3) / (1.0 + 4))
+    end
+
+    it 'returns 0.0 if no text inserted' do
+      old_text = Text.new('deletion text')
+      new_text = Text.new('text')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/vulgarism_frequency_spec.rb

@@ -0,0 +1,30 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::VulgarismFrequency do
+  it { is_expected.to be_a Features::FrequencyBase }
+
+  describe '#calculate' do
+    it 'returns the number of vulgarism words over all inserted words' do
+      # total 8 inserted words, 3 vulgarism
+      old_text = Text.new('Old shit.')
+      new_text = Text.new('Old shit. Fuck, fu*ck you $lut, and all the others.')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 3.0 / 8.0
+    end
+
+    it 'returns 0.0 on emtpy clean text revisions' do
+      old_text = Text.new('Old shit.')
+      new_text = Text.new('Old shit. {{speedy deletion}}')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/vulgarism_impact_spec.rb

@@ -0,0 +1,52 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::VulgarismImpact do
+  it { is_expected.to be_a Features::ImpactBase }
+
+  describe '#calculate' do
+    it 'returns the impact of vulgarism words of the new revision text' do
+      # 1 vulgarism
+      old_text = Text.new('Fuck you, you and all the others')
+      # 3 vulgarism
+      new_text = Text.new('Fuck you, fuck you, you and all the others sluts')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 1.0 / (1.0 + 3.0)
+    end
+
+    it 'returns 0.5 when both revision texts have no terms' do
+      text = Text.new('{{speedy deletion}}')
+
+      old_rev = build(:old_revision, text: text)
+      new_rev = build(:new_revision, text: text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.5
+    end
+
+    it 'returns 0.0 for an emtpy clean text of the old revision' do
+      old_text = Text.new('{{speedy deletion}}')
+      new_text = Text.new('fuck')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 0.0
+    end
+
+    it 'returns 1.0 for an emtpy clean text of the new revision' do
+      old_text = Text.new('fuck')
+      new_text = Text.new('{{speedy deletion}}')
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, new_revision: new_rev, old_revision: old_rev)
+
+      expect(subject.calculate(edit)).to eq 1.0
+    end
+  end
+end

data/spec/vandalism_detection/features/weekday_spec.rb

@@ -0,0 +1,16 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::Weekday do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns the weekday as decimal value' do
+      old_rev = build(:old_revision)
+      new_rev = build(:new_revision, timestamp: '2012-12-11T05:30:36Z')
+
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 2 # Thuesday
+    end
+  end
+end

data/spec/vandalism_detection/features/words_increment_spec.rb

@@ -0,0 +1,29 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::WordsIncrement do
+  it { is_expected.to be_a Features::Base }
+
+  describe '#calculate' do
+    it 'returns a negative increment for more removed texts' do
+      old_text = Text.new('one two three four five six seven eight') # length 8
+      new_text = Text.new('one two three') # length 3
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 3.0 - 8.0
+    end
+
+    it 'returns a positive increment on more removed texts' do
+      old_text = Text.new('one two three') # length 3
+      new_text = Text.new('one two three four five six seven eight') # length 8
+
+      old_rev = build(:old_revision, text: old_text)
+      new_rev = build(:new_revision, text: new_text)
+      edit = build(:edit, old_revision: old_rev, new_revision: new_rev)
+
+      expect(subject.calculate(edit)).to eq 8.0 - 3.0
+    end
+  end
+end