wikipedia-vandalism_detection 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +17 -0
- data/Gemfile +6 -0
- data/LICENSE.txt +4 -0
- data/README.md +265 -0
- data/Rakefile +12 -0
- data/lib/java/LibSVM.jar +0 -0
- data/lib/java/SMOTE.jar +0 -0
- data/lib/java/balancedRandomForest.jar +0 -0
- data/lib/java/diffutils-1.3.0.jar +0 -0
- data/lib/java/libsvm.jar +0 -0
- data/lib/java/oneClassClassifier.jar +0 -0
- data/lib/java/realAdaBoost.jar +0 -0
- data/lib/java/swc-engine-1.1.0-jar-with-dependencies.jar +0 -0
- data/lib/java/sweble-wikitext-extractor.jar +0 -0
- data/lib/weka/classifiers/functions/lib_svm.rb +15 -0
- data/lib/weka/classifiers/meta/one_class_classifier.rb +25 -0
- data/lib/weka/classifiers/meta/real_ada_boost.rb +17 -0
- data/lib/weka/classifiers/trees/balanced_random_forest.rb +18 -0
- data/lib/weka/filters/supervised/instance/smote.rb +22 -0
- data/lib/wikipedia.rb +51 -0
- data/lib/wikipedia/vandalism_detection.rb +30 -0
- data/lib/wikipedia/vandalism_detection/algorithms.rb +18 -0
- data/lib/wikipedia/vandalism_detection/algorithms/kullback_leibler_divergence.rb +69 -0
- data/lib/wikipedia/vandalism_detection/classifier.rb +186 -0
- data/lib/wikipedia/vandalism_detection/configuration.rb +321 -0
- data/lib/wikipedia/vandalism_detection/diff.rb +27 -0
- data/lib/wikipedia/vandalism_detection/edit.rb +75 -0
- data/lib/wikipedia/vandalism_detection/evaluator.rb +606 -0
- data/lib/wikipedia/vandalism_detection/exceptions.rb +40 -0
- data/lib/wikipedia/vandalism_detection/feature_calculator.rb +89 -0
- data/lib/wikipedia/vandalism_detection/features.rb +67 -0
- data/lib/wikipedia/vandalism_detection/features/all_wordlists_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/all_wordlists_impact.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/anonymity.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/anonymity_previous.rb +29 -0
- data/lib/wikipedia/vandalism_detection/features/article_size.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/bad_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/bad_impact.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/base.rb +54 -0
- data/lib/wikipedia/vandalism_detection/features/biased_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/biased_impact.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/blanking.rb +25 -0
- data/lib/wikipedia/vandalism_detection/features/character_diversity.rb +25 -0
- data/lib/wikipedia/vandalism_detection/features/character_sequence.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/comment_bad_frequency.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/comment_biased_frequency.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/comment_length.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/comment_markup_frequency.rb +27 -0
- data/lib/wikipedia/vandalism_detection/features/comment_pronoun_frequency.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/comment_sex_frequency.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/comment_vulgarism_frequency.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/compressibility.rb +27 -0
- data/lib/wikipedia/vandalism_detection/features/contains_base.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/copyedit.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/digit_ratio.rb +24 -0
- data/lib/wikipedia/vandalism_detection/features/edits_per_user.rb +65 -0
- data/lib/wikipedia/vandalism_detection/features/emoticons_frequency.rb +27 -0
- data/lib/wikipedia/vandalism_detection/features/emoticons_impact.rb +29 -0
- data/lib/wikipedia/vandalism_detection/features/frequency_base.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/impact_base.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_character_distribution.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_external_links.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_internal_links.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_size.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_words.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/longest_word.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/markup_frequency.rb +27 -0
- data/lib/wikipedia/vandalism_detection/features/markup_impact.rb +27 -0
- data/lib/wikipedia/vandalism_detection/features/non_alphanumeric_ratio.rb +24 -0
- data/lib/wikipedia/vandalism_detection/features/personal_life.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/pronoun_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/pronoun_impact.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_all_wordlists_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/removed_bad_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/removed_biased_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/removed_character_distribution.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/removed_emoticons_frequency.rb +27 -0
- data/lib/wikipedia/vandalism_detection/features/removed_markup_frequency.rb +28 -0
- data/lib/wikipedia/vandalism_detection/features/removed_pronoun_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/removed_sex_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/removed_size.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/removed_vulgarism_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/removed_words.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/replacement_similarity.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/reverted.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/revisions_character_distribution.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/same_editor.rb +29 -0
- data/lib/wikipedia/vandalism_detection/features/sex_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/sex_impact.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/size_increment.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/size_ratio.rb +26 -0
- data/lib/wikipedia/vandalism_detection/features/term_frequency.rb +25 -0
- data/lib/wikipedia/vandalism_detection/features/time_interval.rb +31 -0
- data/lib/wikipedia/vandalism_detection/features/time_of_day.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/upper_case_ratio.rb +24 -0
- data/lib/wikipedia/vandalism_detection/features/upper_case_words_ratio.rb +31 -0
- data/lib/wikipedia/vandalism_detection/features/upper_to_lower_case_ratio.rb +24 -0
- data/lib/wikipedia/vandalism_detection/features/user_reputation.rb +38 -0
- data/lib/wikipedia/vandalism_detection/features/vulgarism_frequency.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/vulgarism_impact.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/weekday.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/words_increment.rb +22 -0
- data/lib/wikipedia/vandalism_detection/instances.rb +130 -0
- data/lib/wikipedia/vandalism_detection/page.rb +88 -0
- data/lib/wikipedia/vandalism_detection/page_parser.rb +52 -0
- data/lib/wikipedia/vandalism_detection/revision.rb +69 -0
- data/lib/wikipedia/vandalism_detection/revision_parser.rb +43 -0
- data/lib/wikipedia/vandalism_detection/test_dataset.rb +367 -0
- data/lib/wikipedia/vandalism_detection/text.rb +18 -0
- data/lib/wikipedia/vandalism_detection/training_dataset.rb +303 -0
- data/lib/wikipedia/vandalism_detection/version.rb +5 -0
- data/lib/wikipedia/vandalism_detection/wikitext_extractor.rb +80 -0
- data/lib/wikipedia/vandalism_detection/word_lists.rb +19 -0
- data/lib/wikipedia/vandalism_detection/word_lists/bad.rb +12 -0
- data/lib/wikipedia/vandalism_detection/word_lists/biased.rb +21 -0
- data/lib/wikipedia/vandalism_detection/word_lists/emoticons.rb +22 -0
- data/lib/wikipedia/vandalism_detection/word_lists/markup.rb +12 -0
- data/lib/wikipedia/vandalism_detection/word_lists/pronouns.rb +15 -0
- data/lib/wikipedia/vandalism_detection/word_lists/sex.rb +12 -0
- data/lib/wikipedia/vandalism_detection/word_lists/vulgarism.rb +97 -0
- data/spec/factories/edit.rb +20 -0
- data/spec/factories/page.rb +13 -0
- data/spec/factories/revision.rb +51 -0
- data/spec/resources/config/config.yml +35 -0
- data/spec/resources/corpora/test/edits.csv +8 -0
- data/spec/resources/corpora/test/ground-truth.txt +3 -0
- data/spec/resources/corpora/test/revisions/part-1/326471754.txt +199 -0
- data/spec/resources/corpora/test/revisions/part-1/326873205.txt +137 -0
- data/spec/resources/corpora/test/revisions/part-1/328774035.txt +162 -0
- data/spec/resources/corpora/test/revisions/part-2/307084144.txt +137 -0
- data/spec/resources/corpora/test/revisions/part-2/326978767.txt +199 -0
- data/spec/resources/corpora/test/revisions/part-2/328774110.txt +162 -0
- data/spec/resources/corpora/training/annotations.csv +7 -0
- data/spec/resources/corpora/training/edits.csv +7 -0
- data/spec/resources/corpora/training/revisions/part-1/326471754.txt +199 -0
- data/spec/resources/corpora/training/revisions/part-1/326873205.txt +137 -0
- data/spec/resources/corpora/training/revisions/part-1/328774035.txt +162 -0
- data/spec/resources/corpora/training/revisions/part-2/307084144.txt +137 -0
- data/spec/resources/corpora/training/revisions/part-2/326978767.txt +199 -0
- data/spec/resources/corpora/training/revisions/part-2/328774110.txt +162 -0
- data/spec/resources/page_with_redirects.xml +85 -0
- data/spec/resources/redirect_page.xml +59 -0
- data/spec/resources/revision_simplified.xml +13 -0
- data/spec/resources/sample_revision.txt +137 -0
- data/spec/resources/sample_revision_clean_text.txt +1 -0
- data/spec/resources/sample_revision_plain_text.txt +183 -0
- data/spec/resources/vandalism_on_wikipedia.xml +234 -0
- data/spec/resources/vandalism_on_wikipedia_simplified.xml +119 -0
- data/spec/resources/wikipedia_tokens.txt +30 -0
- data/spec/spec_helper.rb +22 -0
- data/spec/support/macros/file_reading.rb +7 -0
- data/spec/support/macros/test_configuration.rb +71 -0
- data/spec/vandalism_detection/algorithms/kullback_leibler_divergence_spec.rb +36 -0
- data/spec/vandalism_detection/classifier_spec.rb +317 -0
- data/spec/vandalism_detection/configuration_spec.rb +517 -0
- data/spec/vandalism_detection/diff_spec.rb +40 -0
- data/spec/vandalism_detection/edit_spec.rb +137 -0
- data/spec/vandalism_detection/evaluator_spec.rb +671 -0
- data/spec/vandalism_detection/feature_calculator_spec.rb +128 -0
- data/spec/vandalism_detection/features/all_wordlists_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/all_wordlists_impact_spec.rb +58 -0
- data/spec/vandalism_detection/features/anonymity_previous_spec.rb +61 -0
- data/spec/vandalism_detection/features/anonymity_spec.rb +23 -0
- data/spec/vandalism_detection/features/article_size_spec.rb +35 -0
- data/spec/vandalism_detection/features/bad_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/bad_impact_spec.rb +59 -0
- data/spec/vandalism_detection/features/base_spec.rb +49 -0
- data/spec/vandalism_detection/features/biased_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/biased_impact_spec.rb +58 -0
- data/spec/vandalism_detection/features/blanking_spec.rb +38 -0
- data/spec/vandalism_detection/features/character_diversity_spec.rb +35 -0
- data/spec/vandalism_detection/features/character_sequence_spec.rb +37 -0
- data/spec/vandalism_detection/features/comment_bad_frequency_spec.rb +34 -0
- data/spec/vandalism_detection/features/comment_biased_frequency_spec.rb +34 -0
- data/spec/vandalism_detection/features/comment_length_spec.rb +27 -0
- data/spec/vandalism_detection/features/comment_markup_frequency_spec.rb +34 -0
- data/spec/vandalism_detection/features/comment_pronoun_frequency_spec.rb +34 -0
- data/spec/vandalism_detection/features/comment_sex_frequency_spec.rb +34 -0
- data/spec/vandalism_detection/features/comment_vulgarism_frequency_spec.rb +34 -0
- data/spec/vandalism_detection/features/compressibility_spec.rb +42 -0
- data/spec/vandalism_detection/features/contains_base_spec.rb +33 -0
- data/spec/vandalism_detection/features/copyedit_spec.rb +33 -0
- data/spec/vandalism_detection/features/digit_ratio_spec.rb +35 -0
- data/spec/vandalism_detection/features/edits_per_user_spec.rb +49 -0
- data/spec/vandalism_detection/features/emoticons_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/emoticons_impact_spec.rb +51 -0
- data/spec/vandalism_detection/features/frequency_base_spec.rb +26 -0
- data/spec/vandalism_detection/features/impact_base_spec.rb +41 -0
- data/spec/vandalism_detection/features/inserted_character_distribution_spec.rb +46 -0
- data/spec/vandalism_detection/features/inserted_external_links_spec.rb +35 -0
- data/spec/vandalism_detection/features/inserted_internal_links_spec.rb +35 -0
- data/spec/vandalism_detection/features/inserted_size_spec.rb +35 -0
- data/spec/vandalism_detection/features/inserted_words_spec.rb +35 -0
- data/spec/vandalism_detection/features/longest_word_spec.rb +35 -0
- data/spec/vandalism_detection/features/markup_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/markup_impact_spec.rb +59 -0
- data/spec/vandalism_detection/features/non_alphanumeric_ratio_spec.rb +35 -0
- data/spec/vandalism_detection/features/personal_life_spec.rb +26 -0
- data/spec/vandalism_detection/features/pronoun_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/pronoun_impact_spec.rb +59 -0
- data/spec/vandalism_detection/features/removed_all_wordlists_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/removed_bad_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/removed_biased_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/removed_character_distribution_spec.rb +46 -0
- data/spec/vandalism_detection/features/removed_emoticons_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/removed_markup_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/removed_pronoun_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/removed_sex_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/removed_size_spec.rb +35 -0
- data/spec/vandalism_detection/features/removed_vulgarism_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/removed_words_spec.rb +35 -0
- data/spec/vandalism_detection/features/replacement_similarity_spec.rb +44 -0
- data/spec/vandalism_detection/features/reverted_spec.rb +28 -0
- data/spec/vandalism_detection/features/revisions_character_distribution_spec.rb +46 -0
- data/spec/vandalism_detection/features/same_editor_spec.rb +60 -0
- data/spec/vandalism_detection/features/sex_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/sex_impact_spec.rb +59 -0
- data/spec/vandalism_detection/features/size_increment_spec.rb +35 -0
- data/spec/vandalism_detection/features/size_ratio_spec.rb +57 -0
- data/spec/vandalism_detection/features/term_frequency_spec.rb +38 -0
- data/spec/vandalism_detection/features/time_interval_spec.rb +50 -0
- data/spec/vandalism_detection/features/time_of_day_spec.rb +22 -0
- data/spec/vandalism_detection/features/upper_case_ratio_spec.rb +35 -0
- data/spec/vandalism_detection/features/upper_case_words_ratio_spec.rb +37 -0
- data/spec/vandalism_detection/features/upper_to_lower_case_ratio_spec.rb +35 -0
- data/spec/vandalism_detection/features/user_reputation_spec.rb +52 -0
- data/spec/vandalism_detection/features/vulgarism_frequency_spec.rb +36 -0
- data/spec/vandalism_detection/features/vulgarism_impact_spec.rb +58 -0
- data/spec/vandalism_detection/features/weekday_spec.rb +22 -0
- data/spec/vandalism_detection/features/words_increment_spec.rb +35 -0
- data/spec/vandalism_detection/instances_spec.rb +156 -0
- data/spec/vandalism_detection/page_parser_spec.rb +184 -0
- data/spec/vandalism_detection/page_spec.rb +135 -0
- data/spec/vandalism_detection/revision_parser_spec.rb +53 -0
- data/spec/vandalism_detection/revision_spec.rb +115 -0
- data/spec/vandalism_detection/test_dataset_spec.rb +231 -0
- data/spec/vandalism_detection/text_spec.rb +29 -0
- data/spec/vandalism_detection/training_dataset_spec.rb +264 -0
- data/spec/vandalism_detection/wikitext_extractor_spec.rb +72 -0
- data/spec/weka/classifiers/functions/lib_svm_spec.rb +38 -0
- data/spec/weka/classifiers/meta/one_class_classifier_spec.rb +76 -0
- data/spec/weka/classifiers/meta/real_ada_boost_spec.rb +40 -0
- data/spec/weka/classifiers/trees/balanced_random_forest_spec.rb +40 -0
- data/spec/weka/filters/supervised/instance/smote_spec.rb +6 -0
- data/wikipedia-vandalism_detection.gemspec +30 -0
- metadata +512 -0
@@ -0,0 +1,31 @@
|
|
1
|
+
# encoding: UTF-8
|
2
|
+
|
3
|
+
require 'wikipedia/vandalism_detection/features/base'
|
4
|
+
require 'wikipedia/vandalism_detection/text'
|
5
|
+
|
6
|
+
module Wikipedia
|
7
|
+
module VandalismDetection
|
8
|
+
module Features
|
9
|
+
|
10
|
+
# This feature computes the uppercase to all words ratio of the edit's new revision inserted text.
|
11
|
+
class UpperCaseWordsRatio < Base
|
12
|
+
|
13
|
+
def calculate(edit)
|
14
|
+
super
|
15
|
+
|
16
|
+
inserted_alpha_text = edit.inserted_words.delete_if{ |w| w.gsub(/[^A-Za-z]/, '').empty? }.join("\n")
|
17
|
+
words = Text.new(inserted_alpha_text).clean.gsub(/[^\w\s]/, '').split
|
18
|
+
|
19
|
+
return 0.0 if words.empty?
|
20
|
+
|
21
|
+
uppercase_words_count = words.reduce(0) do |count, word|
|
22
|
+
count += 1 if word == word.upcase
|
23
|
+
count
|
24
|
+
end
|
25
|
+
|
26
|
+
(1.0 + uppercase_words_count) / (1.0 + words.count)
|
27
|
+
end
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
31
|
+
end
|
@@ -0,0 +1,24 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
|
7
|
+
# This feature computes the uppercase to all letters ratio of the edit's new revision inserted text.
|
8
|
+
class UpperToLowerCaseRatio< Base
|
9
|
+
|
10
|
+
def calculate(edit)
|
11
|
+
super
|
12
|
+
|
13
|
+
text = edit.inserted_text
|
14
|
+
return 0.0 if text.empty?
|
15
|
+
|
16
|
+
uppercase_count = text.scan(/[[:upper:]]/).size
|
17
|
+
lowercase_count = text.scan(/[[:lower:]]/).size
|
18
|
+
|
19
|
+
(1.0 + uppercase_count) / (1.0 + lowercase_count)
|
20
|
+
end
|
21
|
+
end
|
22
|
+
end
|
23
|
+
end
|
24
|
+
end
|
@@ -0,0 +1,38 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
require 'open-uri'
|
3
|
+
require 'nokogiri'
|
4
|
+
|
5
|
+
module Wikipedia
|
6
|
+
module VandalismDetection
|
7
|
+
module Features
|
8
|
+
|
9
|
+
# This feature calculates the average editor's reputation on the current article.
|
10
|
+
class UserReputation < Base
|
11
|
+
|
12
|
+
# Attention! This can be pretty time consuming (up to 2 sec) due to the url request.
|
13
|
+
def calculate(edit)
|
14
|
+
super
|
15
|
+
|
16
|
+
revision = edit.new_revision
|
17
|
+
page_id = edit.page.id || Wikipedia::api_request({ titles: edit.page.title }).xpath("//page/@pageid").first
|
18
|
+
|
19
|
+
text = Wikipedia::wikitrust_request({ pageid: page_id, revid: revision.id })
|
20
|
+
contributions = text.scan(/(\{\{#t:\d+,\d+,#{revision.contributor}\}\})/)
|
21
|
+
# {{#t:trust,revision_id,UserName}}
|
22
|
+
|
23
|
+
trust = 0.0
|
24
|
+
|
25
|
+
unless contributions.empty?
|
26
|
+
sum = contributions.reduce(0.0) do |sum, contribution|
|
27
|
+
sum += contribution[0].split(',').first.split(':').last.to_f
|
28
|
+
end
|
29
|
+
|
30
|
+
trust = sum / contributions.count.to_f
|
31
|
+
end
|
32
|
+
|
33
|
+
trust
|
34
|
+
end
|
35
|
+
end
|
36
|
+
end
|
37
|
+
end
|
38
|
+
end
|
@@ -0,0 +1,23 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/frequency_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/vulgarism'
|
3
|
+
require 'wikipedia/vandalism_detection/text'
|
4
|
+
|
5
|
+
module Wikipedia
|
6
|
+
module VandalismDetection
|
7
|
+
module Features
|
8
|
+
|
9
|
+
# This feature computes frequency of vulgarism words in the inserted text.
|
10
|
+
class VulgarismFrequency < FrequencyBase
|
11
|
+
|
12
|
+
# Returns the percentage of vulgarism words in the inserted text.
|
13
|
+
# Returns 0.0 if inserted clean text is of zero length.
|
14
|
+
def calculate(edit)
|
15
|
+
super
|
16
|
+
|
17
|
+
text = Text.new(edit.inserted_words.join("\n")).clean
|
18
|
+
frequency(text, WordLists::VULGARISM)
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
22
|
+
end
|
23
|
+
end
|
@@ -0,0 +1,22 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/impact_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/vulgarism'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
|
8
|
+
# This feature computes the percentage by which the edit increases the number of vulgarism words in the text.
|
9
|
+
class VulgarismImpact < ImpactBase
|
10
|
+
|
11
|
+
def calculate(edit)
|
12
|
+
super
|
13
|
+
|
14
|
+
old_text = edit.old_revision.text.clean
|
15
|
+
new_text = edit.new_revision.text.clean
|
16
|
+
|
17
|
+
impact(old_text, new_text, WordLists::VULGARISM)
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
22
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
require 'date'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
|
8
|
+
# This feature calculates the weekday of new revision edit as numeric value.
|
9
|
+
# Monday => 1, Thuesday => 2, etc.
|
10
|
+
class Weekday < Base
|
11
|
+
|
12
|
+
def calculate(edit)
|
13
|
+
super
|
14
|
+
|
15
|
+
timestamp = edit.new_revision.timestamp
|
16
|
+
DateTime.parse(timestamp).wday
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,22 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
|
7
|
+
# This feature computes the increment of the edit's revisions words.
|
8
|
+
class WordsIncrement < Base
|
9
|
+
|
10
|
+
# computation: |inserted| - |removed|
|
11
|
+
def calculate(edit)
|
12
|
+
super
|
13
|
+
|
14
|
+
inserted_count = edit.inserted_words.count
|
15
|
+
removed_count = edit.removed_words.count
|
16
|
+
|
17
|
+
inserted_count - removed_count
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
22
|
+
end
|
@@ -0,0 +1,130 @@
|
|
1
|
+
require 'ruby-band'
|
2
|
+
require 'wikipedia/vandalism_detection/configuration'
|
3
|
+
require 'weka/classifiers/meta/one_class_classifier'
|
4
|
+
|
5
|
+
module Wikipedia
|
6
|
+
module VandalismDetection
|
7
|
+
|
8
|
+
class Instances
|
9
|
+
|
10
|
+
REGULAR_CLASS_INDEX = 0
|
11
|
+
VANDALISM_CLASS_INDEX = 1
|
12
|
+
NOT_KNOWN_INDEX = 2
|
13
|
+
|
14
|
+
CLASS = 'class'
|
15
|
+
VANDALISM = 'vandalism'
|
16
|
+
REGULAR = 'regular'
|
17
|
+
NOT_KNOWN = '?'
|
18
|
+
OUTLIER = Weka::Classifiers::Meta::OneClassClassifier::OUTLIER_LABEL
|
19
|
+
|
20
|
+
VANDALISM_SHORT= 'V'
|
21
|
+
REGULAR_SHORT = 'R'
|
22
|
+
|
23
|
+
OLD_REVISION_ID = 'oldrevisionid'
|
24
|
+
NEW_REVISION_ID = 'newrevisionid'
|
25
|
+
|
26
|
+
|
27
|
+
CLASSES = {
|
28
|
+
REGULAR_CLASS_INDEX => REGULAR,
|
29
|
+
VANDALISM_CLASS_INDEX => VANDALISM,
|
30
|
+
NOT_KNOWN_INDEX => NOT_KNOWN
|
31
|
+
}
|
32
|
+
|
33
|
+
CLASSES_SHORT = {
|
34
|
+
REGULAR_CLASS_INDEX => REGULAR_SHORT,
|
35
|
+
VANDALISM_CLASS_INDEX => VANDALISM_SHORT,
|
36
|
+
NOT_KNOWN_INDEX => NOT_KNOWN
|
37
|
+
}
|
38
|
+
|
39
|
+
# Returns an empty instances dataset of type Java::WekaCore::Instances::Base.
|
40
|
+
# This dataset is used for feature computation and classification for Wikipedia vandalism detection
|
41
|
+
# while training.
|
42
|
+
#
|
43
|
+
# @example
|
44
|
+
# datset = Wikipedia::VandalismDetection::Instances.empty
|
45
|
+
# => #<Java::WekaCore::Instances::Base:0xf0f9a00
|
46
|
+
# @positions=[
|
47
|
+
# #<Java::WekaCore::Attribute:0x17207a76>,
|
48
|
+
# #<Java::WekaCore::Attribute:0x5547e4d6>,
|
49
|
+
# #<Java::WekaCore::Attribute:0x6300c957>,
|
50
|
+
# ...,
|
51
|
+
# #<Java::WekaCore::Attribute:0x5a74fae4>]>
|
52
|
+
def self.empty
|
53
|
+
features = Wikipedia::VandalismDetection.configuration.features
|
54
|
+
classes = dataset_classes
|
55
|
+
|
56
|
+
dataset = Core::Type::Instances::Base.new do
|
57
|
+
features.each do |name|
|
58
|
+
numeric :"#{name.gsub(' ', '_')}"
|
59
|
+
end
|
60
|
+
|
61
|
+
nominal :class, classes
|
62
|
+
end
|
63
|
+
|
64
|
+
dataset.class_index = features.count
|
65
|
+
dataset
|
66
|
+
end
|
67
|
+
|
68
|
+
# Returns an empty instances dataset of type Java::WekaCore::Instances::Base.
|
69
|
+
# This dataset is used for feature computation and classification for Wikipedia vandalism detection
|
70
|
+
# while training.
|
71
|
+
#
|
72
|
+
# @example
|
73
|
+
# datset = Wikipedia::VandalismDetection::Instances.empty
|
74
|
+
# => #<Java::WekaCore::Instances::Base:0xf0f9a00
|
75
|
+
# @positions=[
|
76
|
+
# #<Java::WekaCore::Attribute:0x17207a76>
|
77
|
+
def self.empty_for_feature(name)
|
78
|
+
classes = dataset_classes
|
79
|
+
|
80
|
+
dataset = Core::Type::Instances::Base.new do
|
81
|
+
numeric :"#{name.gsub(' ', '_')}"
|
82
|
+
nominal :class, classes
|
83
|
+
end
|
84
|
+
|
85
|
+
dataset.class_index = 1
|
86
|
+
dataset
|
87
|
+
end
|
88
|
+
|
89
|
+
# Returns an empty instances dataset of type Java::WekaCore::Instances::Base.
|
90
|
+
# This dataset is used for feature computation and classification for Wikipedia vandalism detection
|
91
|
+
# while testing.
|
92
|
+
#
|
93
|
+
# @example
|
94
|
+
# datset = Wikipedia::VandalismDetection::Instances.empty_for_test
|
95
|
+
# => #<Java::WekaCore::Instances::Base:0xf0f9a00
|
96
|
+
# @positions=[
|
97
|
+
# #<Java::WekaCore::Attribute:0x17207a76>]>
|
98
|
+
def self.empty_for_test_feature(name)
|
99
|
+
dataset = Core::Type::Instances::Base.new do
|
100
|
+
numeric :"#{name.gsub(' ', '_')}"
|
101
|
+
numeric OLD_REVISION_ID.to_sym
|
102
|
+
numeric NEW_REVISION_ID.to_sym
|
103
|
+
end
|
104
|
+
|
105
|
+
dataset
|
106
|
+
end
|
107
|
+
|
108
|
+
# Returns an empty instances dataset of type Java::WekaCore::Instances::Base.
|
109
|
+
# This dataset is used for creating the ground truth classification.
|
110
|
+
def self.empty_for_test_class
|
111
|
+
classes = dataset_classes
|
112
|
+
|
113
|
+
dataset = Core::Type::Instances::Base.new do
|
114
|
+
nominal :class, classes
|
115
|
+
end
|
116
|
+
|
117
|
+
dataset
|
118
|
+
end
|
119
|
+
|
120
|
+
def self.dataset_classes
|
121
|
+
classes = Array.new
|
122
|
+
classes[VANDALISM_CLASS_INDEX] = VANDALISM
|
123
|
+
classes[REGULAR_CLASS_INDEX] = REGULAR
|
124
|
+
classes
|
125
|
+
end
|
126
|
+
|
127
|
+
private_class_method :dataset_classes
|
128
|
+
end
|
129
|
+
end
|
130
|
+
end
|
@@ -0,0 +1,88 @@
|
|
1
|
+
require 'nokogiri'
|
2
|
+
require_relative 'edit'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
|
7
|
+
# Represents a full wikipedia page history.
|
8
|
+
class Page
|
9
|
+
|
10
|
+
START_TAG = '<page>'.freeze
|
11
|
+
END_TAG = '</page>'.freeze
|
12
|
+
|
13
|
+
attr_accessor :id, :title
|
14
|
+
attr_reader :revisions
|
15
|
+
|
16
|
+
def initialize
|
17
|
+
@revisions = {}
|
18
|
+
@edits = []
|
19
|
+
@update_edits = false
|
20
|
+
@update_reverted_edits = false
|
21
|
+
end
|
22
|
+
|
23
|
+
def add_revision(revision)
|
24
|
+
@revisions[revision.id] = revision
|
25
|
+
|
26
|
+
@update_edits = true
|
27
|
+
@update_reverted_edits = true
|
28
|
+
end
|
29
|
+
|
30
|
+
def edits
|
31
|
+
@edits = create_edits_from @revisions if @update_edits
|
32
|
+
@edits
|
33
|
+
end
|
34
|
+
|
35
|
+
def reverted_edits
|
36
|
+
@reverted_edits = create_reverted_edits_from @revisions if @update_reverted_edits
|
37
|
+
@reverted_edits
|
38
|
+
end
|
39
|
+
|
40
|
+
private
|
41
|
+
|
42
|
+
def create_edits_from(revisions)
|
43
|
+
@update_edits = false
|
44
|
+
edits = []
|
45
|
+
|
46
|
+
revisions.each do |_, new_revision|
|
47
|
+
old_revision = revisions[new_revision.parent_id]
|
48
|
+
edits << Edit.new(old_revision, new_revision) unless old_revision.nil?
|
49
|
+
end
|
50
|
+
|
51
|
+
edits.each { |edit| edit.instance_variable_set(:@page, self) }
|
52
|
+
edits
|
53
|
+
end
|
54
|
+
|
55
|
+
# Returns the reverted edits by comparing the text's sha1 hashes of multiple revisions.
|
56
|
+
# If the next but one revision has the same sha1 hash as a base revision and the base revision has another
|
57
|
+
# hash than the one before, the in-between revision is reverted.
|
58
|
+
# The resulting edit holds the base revision as old revision and the reverted as new revision.
|
59
|
+
def create_reverted_edits_from(revisions)
|
60
|
+
@update_reverted_edits = false
|
61
|
+
edits = []
|
62
|
+
|
63
|
+
revisions.each do |current_id, first_revision|
|
64
|
+
second_revision_select = revisions.select { |_, value| value.parent_id == current_id }.first
|
65
|
+
next unless second_revision_select
|
66
|
+
|
67
|
+
second_revision = second_revision_select[1]
|
68
|
+
|
69
|
+
third_revision_select = revisions.select { |_, value| value.parent_id == second_revision.id }.first
|
70
|
+
next unless third_revision_select
|
71
|
+
|
72
|
+
first_sha1 = first_revision.sha1
|
73
|
+
second_sha1 = second_revision.sha1
|
74
|
+
third_sha1 = third_revision_select[1].sha1
|
75
|
+
|
76
|
+
previous_revision_select = revisions.select { |_, value| value.id == first_revision.parent_id }.first
|
77
|
+
previous_sha1 = previous_revision_select ? previous_revision_select[1].sha1 : nil
|
78
|
+
|
79
|
+
if (first_sha1 == third_sha1) && (second_sha1 != previous_sha1)
|
80
|
+
edits << Edit.new(first_revision, second_revision)
|
81
|
+
end
|
82
|
+
end
|
83
|
+
|
84
|
+
edits
|
85
|
+
end
|
86
|
+
end
|
87
|
+
end
|
88
|
+
end
|
@@ -0,0 +1,52 @@
|
|
1
|
+
require 'nokogiri'
|
2
|
+
require_relative 'page'
|
3
|
+
require_relative 'revision'
|
4
|
+
|
5
|
+
module Wikipedia
|
6
|
+
module VandalismDetection
|
7
|
+
class PageParser
|
8
|
+
|
9
|
+
# Parses an xml string and returns a Wikipedia::VandalismDetection::Page.
|
10
|
+
def parse(xml)
|
11
|
+
@page = Page.new
|
12
|
+
document = Nokogiri::XML(xml, nil, 'UTF-8')
|
13
|
+
|
14
|
+
@page.title = document.xpath('//page/title').inner_text
|
15
|
+
@page.id = document.xpath('//page/id').inner_text
|
16
|
+
|
17
|
+
build_revisions_from document
|
18
|
+
|
19
|
+
@page
|
20
|
+
end
|
21
|
+
|
22
|
+
private
|
23
|
+
|
24
|
+
# Builds and saves the available revisions to the @page variable
|
25
|
+
def build_revisions_from(document)
|
26
|
+
elements = document.xpath('//revision')
|
27
|
+
|
28
|
+
elements.each do |element|
|
29
|
+
revision = Revision.new
|
30
|
+
|
31
|
+
revision.id = element.xpath('id').inner_text
|
32
|
+
revision.timestamp = element.xpath('timestamp').inner_text
|
33
|
+
revision.comment = Text.new(element.xpath('comment').inner_text)
|
34
|
+
revision.text = Text.new(element.xpath('text').inner_text)
|
35
|
+
revision.sha1 = element.xpath('sha1').inner_text
|
36
|
+
|
37
|
+
parent_id_node = element.xpath('parentid')
|
38
|
+
contributor_id_node = element.xpath('contributor/id')
|
39
|
+
contributor_ip_node = element.xpath('contributor/ip')
|
40
|
+
contributor_username_node = element.xpath('contributor/username')
|
41
|
+
|
42
|
+
revision.parent_id = parent_id_node.inner_text unless parent_id_node.empty?
|
43
|
+
revision.contributor = contributor_id_node.inner_text unless contributor_id_node.empty?
|
44
|
+
revision.contributor = contributor_ip_node.inner_text unless contributor_ip_node.empty?
|
45
|
+
revision.contributor_username = contributor_username_node.inner_text unless contributor_username_node.empty?
|
46
|
+
|
47
|
+
@page.add_revision(revision)
|
48
|
+
end
|
49
|
+
end
|
50
|
+
end
|
51
|
+
end
|
52
|
+
end
|