wikipedia-vandalism_detection 0.0.1

data/spec/vandalism_detection/features/sex_frequency_spec.rb

@@ -0,0 +1,36 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SexFrequency do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::SexFrequency.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::FrequencyBase }
+
+  describe "#calculate" do
+
+    it "returns the number of vulgarism words relative to all words count" do
+      # total 6 words, 3 bad
+      old_text = Wikipedia::VandalismDetection::Text.new 'Old whatever.'
+      new_text = Wikipedia::VandalismDetection::Text.new "Old whatever. New sex contents. Penis, dildos, boy."
+
+      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 / 6.0
+    end
+
+    it "returns 0.0 on emtpy clean text revisions" do
+      old_text = Wikipedia::VandalismDetection::Text.new 'Old guy.'
+      new_text = Wikipedia::VandalismDetection::Text.new "Old guy. {{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/sex_impact_spec.rb

@@ -0,0 +1,59 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SexImpact do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::SexImpact.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::ImpactBase }
+
+  describe "#calculate" do
+
+    it "returns the impact of sex words of the edit's new revision text" do
+      # 3 sex words
+      old_text = Wikipedia::VandalismDetection::Text.new "Penis, old text dildo, breast it"
+
+      # 4 sex words
+      new_text = Wikipedia::VandalismDetection::Text.new "Penis, old text dildo, breast anal it"
+
+      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 / (3.0 + 4.0)
+    end
+
+    it "returns 0.5 on both no terms in text revisions" do
+      text = Wikipedia::VandalismDetection::Text.new "{{speedy deletion}}"
+
+      old_revision = build(:old_revision, text: text)
+      new_revision = build(:new_revision, text: text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.5
+    end
+
+    it "returns 0.0 on emtpy clean text of old revision" do
+      old_text = Wikipedia::VandalismDetection::Text.new "{{speedy deletion}}"
+      new_text = Wikipedia::VandalismDetection::Text.new "anal"
+
+      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 0.0
+    end
+
+    it "returns 1.0 on emtpy clean text of new revision" do
+      old_text = Wikipedia::VandalismDetection::Text.new "anal"
+      new_text = Wikipedia::VandalismDetection::Text.new "{{speedy deletion}}"
+
+      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 1.0
+    end
+  end
+end

data/spec/vandalism_detection/features/size_increment_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SizeIncrement do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::SizeIncrement.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns a negative increment on more removed texts" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new '123456789' # length 9
+      new_revision_text = Wikipedia::VandalismDetection::Text.new '123' # length 3
+
+      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 3.0 - 9.0
+    end
+
+    it "returns a positive increment on more removed texts" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new '123' # length 3
+      new_revision_text = Wikipedia::VandalismDetection::Text.new '123456789' # length 9
+
+      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 9.0 - 3.0
+    end
+  end
+end

data/spec/vandalism_detection/features/size_ratio_spec.rb

@@ -0,0 +1,57 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::SizeRatio do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::SizeRatio.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the ratio of the edit's revisions text sizes" do
+      old_revision_text = Wikipedia::VandalismDetection::Text.new '123456789' # length 9
+      new_revision_text = Wikipedia::VandalismDetection::Text.new '123' # length 3
+
+      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 9.0 / (9.0 + 3.0)
+    end
+
+    it "returns 1.0 on emtpy new text revisions" do
+      old_text = Wikipedia::VandalismDetection::Text.new "sample 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 1.0
+    end
+
+    it "returns 0.0 on emtpy old text revisions" do
+      old_text = Wikipedia::VandalismDetection::Text.new ""
+      new_text = Wikipedia::VandalismDetection::Text.new "sample 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 0.0
+    end
+
+    it "returns 0.5 on emtpy both text revisions" do
+      old_text = Wikipedia::VandalismDetection::Text.new ""
+      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 0.5
+    end
+  end
+end

data/spec/vandalism_detection/features/term_frequency_spec.rb

@@ -0,0 +1,38 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::TermFrequency do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::TermFrequency.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::FrequencyBase }
+
+  describe "#calculate" do
+
+    it "returns the average relative frequency of inserted words in the new revision" 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
+      # avrg = (6/10 + 3/10)/2
+      old_text = Wikipedia::VandalismDetection::Text.new "we\nwe\nyou\nyou\nyou\nI\nyou\nI\n"
+      new_text = Wikipedia::VandalismDetection::Text.new "we\nwe\nyou\nyou\nI\nyou\n''(you''\nyou\nyou\n[[we]])\n"
+
+      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 ((6.0/10.0) + (3.0/10.0)) / 2.0
+    end
+
+    it "returns 0.0 on emtpy clean text revisions" do
+      text = Wikipedia::VandalismDetection::Text.new "{{speedy deletion}}"
+
+      old_revision = build(:old_revision, text: text)
+      new_revision = build(:new_revision, text: text)
+      edit = build(:edit, new_revision: new_revision, old_revision: old_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/time_interval_spec.rb

@@ -0,0 +1,50 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::TimeInterval do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::TimeInterval.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns time interval in days from old to new revision" do
+      old_revision_time = '2014-11-27T18:00:00Z'
+      new_revision_time = '2014-11-29T06:00:00Z'
+
+      old_revision = build(:old_revision, timestamp: old_revision_time)
+      new_revision = build(:new_revision, timestamp: new_revision_time)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 1.5
+    end
+
+    it "requests the time from Wikipedia API if old revisions timestamp is not given" do
+      # to get api call, see:
+      # https://en.wikipedia.org/w/api.php?action=query&prop=revisions&rvprop=timestamp&revids=327585467
+      # => 2009-11-24T01:57:35Z
+      new_revision_time = '2009-11-24T13:57:35Z'
+
+      old_revision = build(:old_revision, id: '327585467', timestamp: nil)
+      new_revision = build(:new_revision, id: '327607921', parent_id: '327585467', timestamp: new_revision_time)
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 0.5
+    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"/>
+      new_revision_time = '2011-11-11T01:00:00Z'
+
+      old_revision = build(:old_revision, id: '325218985', timestamp: nil)
+      new_revision = build(:new_revision, id: '326980599', parent_id: '325218985', timestamp: new_revision_time)
+      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

data/spec/vandalism_detection/features/time_of_day_spec.rb

@@ -0,0 +1,22 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::TimeOfDay do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::TimeOfDay.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the time of day as decimal value (hours)" do
+      old_revision = build(:old_revision)
+      new_revision = build(:new_revision, timestamp: '2012-12-09T05:30:36Z' )
+
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision)
+
+      expect(@feature.calculate(edit)).to eq 5.6
+    end
+  end
+end

data/spec/vandalism_detection/features/upper_case_ratio_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::UpperCaseRatio do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::UpperCaseRatio.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the uppercase to all letters ratio of the edit's new revision cleaned inserted text" do
+      old_text = Wikipedia::VandalismDetection::Text.new('text')
+      new_text = Wikipedia::VandalismDetection::Text.new 'text [[1A 4B6 8Cd]]' # 3 uppercase letters of total 4 inserted letters
+
+      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 (1.0 + 3) / (1.0 + 4)
+    end
+
+    it "returns 0.0 if no text inserted" do
+      old_text = Wikipedia::VandalismDetection::Text.new("deletion text")
+      new_text = Wikipedia::VandalismDetection::Text.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 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/upper_case_words_ratio_spec.rb

@@ -0,0 +1,37 @@
+# encoding: UTF-8
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::UpperCaseWordsRatio do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::UpperCaseWordsRatio.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the uppercase to all words ratio of the edit's new revision cleaned inserted text" do
+      old_text = Wikipedia::VandalismDetection::Text.new('text')
+      # 2 two capital (not numbers!) words of total 4 inserted, template {{23A}} is removed while cleaning!
+      new_text = Wikipedia::VandalismDetection::Text.new "text [[HELLO you]] NICE boyß3 1990 {{23A}}"
+
+      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 (1.0 + 2) / (1.0 + 4)
+    end
+
+    it "returns 0.0 if no text inserted" do
+      old_text = Wikipedia::VandalismDetection::Text.new("DELECTION text")
+      new_text = Wikipedia::VandalismDetection::Text.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 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/upper_to_lower_case_ratio_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::UpperToLowerCaseRatio do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::UpperToLowerCaseRatio.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the uppercase to lowercase letters ratio of the edit's new revision inserted text" do
+      old_text = Wikipedia::VandalismDetection::Text.new('text')
+      new_text = Wikipedia::VandalismDetection::Text.new 'text [[1aA 4B6 8Cd ef]]' # 3 uppercase letters, 4 lowercase letters
+
+      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 (1.0 + 3) / (1.0 + 4)
+    end
+
+    it "returns 0.0 if no text inserted" do
+      old_text = Wikipedia::VandalismDetection::Text.new("deletion text")
+      new_text = Wikipedia::VandalismDetection::Text.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 0.0
+    end
+  end
+end

data/spec/vandalism_detection/features/user_reputation_spec.rb

@@ -0,0 +1,52 @@
+require 'spec_helper'
+
+describe Wikipedia::VandalismDetection::Features::UserReputation do
+
+  before do
+    @feature = Wikipedia::VandalismDetection::Features::UserReputation.new
+  end
+
+  it { should be_a Wikipedia::VandalismDetection::Features::Base }
+
+  describe "#calculate" do
+
+    it "returns the average number of contributor's reputation (WikiTrust) in the new revision" do
+      #http://en.collaborativetrust.com/WikiTrust/RemoteAPI?method=wikimarkup&pageid=<page_id>&revid=<old_rev_id>
+      # here: http://en.collaborativetrust.com/WikiTrust/RemoteAPI?method=wikimarkup&pageid=15580374&revid=604074768
+      old_revision = build(:old_revision, id: '603070163')
+      new_revision = build(:new_revision, id: '604074768', parent_id: '603070163', contributor: 'David Levy')
+      page = Wikipedia::VandalismDetection::Page.new
+      page.id = "15580374"
+
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision, page: page)
+
+      expect(@feature.calculate(edit)).to eq 10.0
+    end
+
+    it "fetches the page id from the wikipedia API by page title if not available in edit" do
+      #http://en.collaborativetrust.com/WikiTrust/RemoteAPI?method=wikimarkup&pageid=<page_id>&revid=<old_rev_id>
+      # here: http://en.collaborativetrust.com/WikiTrust/RemoteAPI?method=wikimarkup&pageid=15580374&revid=604074768
+      old_revision = build(:old_revision, id: '603070163')
+      new_revision = build(:new_revision, id: '604074768', parent_id: '603070163', contributor: 'David Levy')
+      page = Wikipedia::VandalismDetection::Page.new
+      page.title = 'Main Page'
+
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision, page: page)
+
+      expect(@feature.calculate(edit)).to eq 10.0
+    end
+
+    it "returns 0.0 WikiTrust if no contributions are available for given user" do
+      #http://en.collaborativetrust.com/WikiTrust/RemoteAPI?method=wikimarkup&pageid=<page_id>&revid=<old_rev_id>
+      # here: http://en.collaborativetrust.com/WikiTrust/RemoteAPI?method=wikimarkup&pageid=15580374&revid=604074768
+      old_revision = build(:old_revision, id: '603070163')
+      new_revision = build(:new_revision, id: '604074768', parent_id: '603070163', contributor: 'Artificial User')
+      page = Wikipedia::VandalismDetection::Page.new
+      page.id = "15580374"
+
+      edit = build(:edit, old_revision: old_revision, new_revision: new_revision, page: page)
+
+      expect(@feature.calculate(edit)).to eq 0.0
+    end
+  end
+end