wikipedia-vandalism_detection 0.1.0-java
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +19 -0
- data/.rspec +2 -0
- data/.rubocop.yml +35 -0
- data/.travis.yml +6 -0
- data/Gemfile +4 -0
- data/LICENSE.txt +4 -0
- data/README.md +288 -0
- data/Rakefile +11 -0
- data/config/wikipedia-vandalism-detection.yml.example +103 -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/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/meta/one_class_classifier.rb +21 -0
- data/lib/weka/classifiers/meta/real_ada_boost.rb +15 -0
- data/lib/weka/classifiers/trees/balanced_random_forest.rb +16 -0
- data/lib/weka/filters/supervised/instance/smote.rb +15 -0
- data/lib/wikipedia/vandalism_detection/algorithms/kullback_leibler_divergence.rb +103 -0
- data/lib/wikipedia/vandalism_detection/algorithms.rb +12 -0
- data/lib/wikipedia/vandalism_detection/classifier.rb +202 -0
- data/lib/wikipedia/vandalism_detection/configuration.rb +350 -0
- data/lib/wikipedia/vandalism_detection/diff.rb +36 -0
- data/lib/wikipedia/vandalism_detection/edit.rb +81 -0
- data/lib/wikipedia/vandalism_detection/evaluator.rb +640 -0
- data/lib/wikipedia/vandalism_detection/exceptions.rb +47 -0
- data/lib/wikipedia/vandalism_detection/feature_calculator.rb +94 -0
- data/lib/wikipedia/vandalism_detection/features/all_wordlists_frequency.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/all_wordlists_impact.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/anonymity.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/anonymity_previous.rb +32 -0
- data/lib/wikipedia/vandalism_detection/features/article_size.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/bad_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/bad_impact.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/base.rb +61 -0
- data/lib/wikipedia/vandalism_detection/features/biased_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/biased_impact.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/blanking.rb +26 -0
- data/lib/wikipedia/vandalism_detection/features/character_diversity.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/character_sequence.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/comment_bad_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/comment_biased_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/comment_length.rb +16 -0
- data/lib/wikipedia/vandalism_detection/features/comment_markup_frequency.rb +30 -0
- data/lib/wikipedia/vandalism_detection/features/comment_pronoun_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/comment_sex_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/comment_vulgarism_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/compressibility.rb +31 -0
- data/lib/wikipedia/vandalism_detection/features/contains_base.rb +16 -0
- data/lib/wikipedia/vandalism_detection/features/copyedit.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/digit_ratio.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/edits_per_user.rb +72 -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 +18 -0
- data/lib/wikipedia/vandalism_detection/features/impact_base.rb +25 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_character_distribution.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_external_links.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_internal_links.rb +18 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_size.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/inserted_words.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/longest_word.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/markup_frequency.rb +29 -0
- data/lib/wikipedia/vandalism_detection/features/markup_impact.rb +30 -0
- data/lib/wikipedia/vandalism_detection/features/non_alphanumeric_ratio.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/personal_life.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/pronoun_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/pronoun_impact.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/removed_all_wordlists_frequency.rb +22 -0
- data/lib/wikipedia/vandalism_detection/features/removed_bad_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_biased_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_character_distribution.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/removed_emoticons_frequency.rb +28 -0
- data/lib/wikipedia/vandalism_detection/features/removed_markup_frequency.rb +30 -0
- data/lib/wikipedia/vandalism_detection/features/removed_pronoun_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_sex_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_size.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/removed_vulgarism_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/removed_words.rb +17 -0
- data/lib/wikipedia/vandalism_detection/features/replacement_similarity.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/reverted.rb +16 -0
- data/lib/wikipedia/vandalism_detection/features/revisions_character_distribution.rb +25 -0
- data/lib/wikipedia/vandalism_detection/features/same_editor.rb +32 -0
- data/lib/wikipedia/vandalism_detection/features/sex_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/sex_impact.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/size_increment.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features/size_ratio.rb +28 -0
- data/lib/wikipedia/vandalism_detection/features/term_frequency.rb +31 -0
- data/lib/wikipedia/vandalism_detection/features/time_interval.rb +38 -0
- data/lib/wikipedia/vandalism_detection/features/time_of_day.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/upper_case_ratio.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/upper_case_words_ratio.rb +33 -0
- data/lib/wikipedia/vandalism_detection/features/upper_to_lower_case_ratio.rb +23 -0
- data/lib/wikipedia/vandalism_detection/features/vulgarism_frequency.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/vulgarism_impact.rb +21 -0
- data/lib/wikipedia/vandalism_detection/features/weekday.rb +19 -0
- data/lib/wikipedia/vandalism_detection/features/words_increment.rb +20 -0
- data/lib/wikipedia/vandalism_detection/features.rb +66 -0
- data/lib/wikipedia/vandalism_detection/instances.rb +121 -0
- data/lib/wikipedia/vandalism_detection/page.rb +101 -0
- data/lib/wikipedia/vandalism_detection/page_parser.rb +63 -0
- data/lib/wikipedia/vandalism_detection/revision.rb +75 -0
- data/lib/wikipedia/vandalism_detection/revision_parser.rb +67 -0
- data/lib/wikipedia/vandalism_detection/test_dataset.rb +374 -0
- data/lib/wikipedia/vandalism_detection/text.rb +23 -0
- data/lib/wikipedia/vandalism_detection/training_dataset.rb +282 -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/bad.rb +11 -0
- data/lib/wikipedia/vandalism_detection/word_lists/biased.rb +20 -0
- data/lib/wikipedia/vandalism_detection/word_lists/emoticons.rb +26 -0
- data/lib/wikipedia/vandalism_detection/word_lists/markup.rb +19 -0
- data/lib/wikipedia/vandalism_detection/word_lists/pronouns.rb +11 -0
- data/lib/wikipedia/vandalism_detection/word_lists/sex.rb +10 -0
- data/lib/wikipedia/vandalism_detection/word_lists/vulgarism.rb +96 -0
- data/lib/wikipedia/vandalism_detection/word_lists.rb +17 -0
- data/lib/wikipedia/vandalism_detection.rb +29 -0
- data/lib/wikipedia.rb +41 -0
- data/spec/factories/edit.rb +19 -0
- data/spec/factories/page.rb +12 -0
- data/spec/factories/revision.rb +51 -0
- data/spec/resources/config/wikipedia-vandalism-detection.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 +38 -0
- data/spec/support/macros/file_reading.rb +6 -0
- data/spec/support/macros/test_configuration.rb +81 -0
- data/spec/vandalism_detection/algorithms/kullback_leibler_divergence_spec.rb +34 -0
- data/spec/vandalism_detection/classifier_spec.rb +330 -0
- data/spec/vandalism_detection/configuration_spec.rb +601 -0
- data/spec/vandalism_detection/diff_spec.rb +40 -0
- data/spec/vandalism_detection/edit_spec.rb +122 -0
- data/spec/vandalism_detection/evaluator_spec.rb +711 -0
- data/spec/vandalism_detection/feature_calculator_spec.rb +135 -0
- data/spec/vandalism_detection/features/all_wordlists_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/all_wordlists_impact_spec.rb +52 -0
- data/spec/vandalism_detection/features/anonymity_previous_spec.rb +67 -0
- data/spec/vandalism_detection/features/anonymity_spec.rb +17 -0
- data/spec/vandalism_detection/features/article_size_spec.rb +29 -0
- data/spec/vandalism_detection/features/bad_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/bad_impact_spec.rb +53 -0
- data/spec/vandalism_detection/features/base_spec.rb +41 -0
- data/spec/vandalism_detection/features/biased_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/biased_impact_spec.rb +52 -0
- data/spec/vandalism_detection/features/blanking_spec.rb +35 -0
- data/spec/vandalism_detection/features/character_diversity_spec.rb +30 -0
- data/spec/vandalism_detection/features/character_sequence_spec.rb +31 -0
- data/spec/vandalism_detection/features/comment_bad_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_biased_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_length_spec.rb +21 -0
- data/spec/vandalism_detection/features/comment_markup_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_pronoun_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_sex_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/comment_vulgarism_frequency_spec.rb +28 -0
- data/spec/vandalism_detection/features/compressibility_spec.rb +38 -0
- data/spec/vandalism_detection/features/contains_base_spec.rb +27 -0
- data/spec/vandalism_detection/features/copyedit_spec.rb +30 -0
- data/spec/vandalism_detection/features/digit_ratio_spec.rb +30 -0
- data/spec/vandalism_detection/features/edits_per_user_spec.rb +51 -0
- data/spec/vandalism_detection/features/emoticons_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/emoticons_impact_spec.rb +45 -0
- data/spec/vandalism_detection/features/frequency_base_spec.rb +20 -0
- data/spec/vandalism_detection/features/impact_base_spec.rb +36 -0
- data/spec/vandalism_detection/features/inserted_character_distribution_spec.rb +40 -0
- data/spec/vandalism_detection/features/inserted_external_links_spec.rb +29 -0
- data/spec/vandalism_detection/features/inserted_internal_links_spec.rb +29 -0
- data/spec/vandalism_detection/features/inserted_size_spec.rb +29 -0
- data/spec/vandalism_detection/features/inserted_words_spec.rb +29 -0
- data/spec/vandalism_detection/features/longest_word_spec.rb +29 -0
- data/spec/vandalism_detection/features/markup_frequency_spec.rb +31 -0
- data/spec/vandalism_detection/features/markup_impact_spec.rb +53 -0
- data/spec/vandalism_detection/features/non_alphanumeric_ratio_spec.rb +30 -0
- data/spec/vandalism_detection/features/personal_life_spec.rb +22 -0
- data/spec/vandalism_detection/features/pronoun_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/pronoun_impact_spec.rb +53 -0
- data/spec/vandalism_detection/features/removed_all_wordlists_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_bad_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_biased_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_character_distribution_spec.rb +40 -0
- data/spec/vandalism_detection/features/removed_emoticons_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_markup_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_pronoun_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_sex_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_size_spec.rb +29 -0
- data/spec/vandalism_detection/features/removed_vulgarism_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/removed_words_spec.rb +29 -0
- data/spec/vandalism_detection/features/replacement_similarity_spec.rb +37 -0
- data/spec/vandalism_detection/features/reverted_spec.rb +24 -0
- data/spec/vandalism_detection/features/revisions_character_distribution_spec.rb +40 -0
- data/spec/vandalism_detection/features/same_editor_spec.rb +71 -0
- data/spec/vandalism_detection/features/sex_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/sex_impact_spec.rb +53 -0
- data/spec/vandalism_detection/features/size_increment_spec.rb +29 -0
- data/spec/vandalism_detection/features/size_ratio_spec.rb +48 -0
- data/spec/vandalism_detection/features/term_frequency_spec.rb +32 -0
- data/spec/vandalism_detection/features/time_interval_spec.rb +56 -0
- data/spec/vandalism_detection/features/time_of_day_spec.rb +16 -0
- data/spec/vandalism_detection/features/upper_case_ratio_spec.rb +30 -0
- data/spec/vandalism_detection/features/upper_case_words_ratio_spec.rb +33 -0
- data/spec/vandalism_detection/features/upper_to_lower_case_ratio_spec.rb +30 -0
- data/spec/vandalism_detection/features/vulgarism_frequency_spec.rb +30 -0
- data/spec/vandalism_detection/features/vulgarism_impact_spec.rb +52 -0
- data/spec/vandalism_detection/features/weekday_spec.rb +16 -0
- data/spec/vandalism_detection/features/words_increment_spec.rb +29 -0
- data/spec/vandalism_detection/instances_spec.rb +146 -0
- data/spec/vandalism_detection/page_parser_spec.rb +190 -0
- data/spec/vandalism_detection/page_spec.rb +134 -0
- data/spec/vandalism_detection/revision_parser_spec.rb +53 -0
- data/spec/vandalism_detection/revision_spec.rb +148 -0
- data/spec/vandalism_detection/test_dataset_spec.rb +227 -0
- data/spec/vandalism_detection/text_spec.rb +29 -0
- data/spec/vandalism_detection/training_dataset_spec.rb +266 -0
- data/spec/vandalism_detection/wikitext_extractor_spec.rb +97 -0
- data/spec/weka/classifiers/meta/one_class_classifier_spec.rb +82 -0
- data/spec/weka/classifiers/meta/real_ada_boost_spec.rb +37 -0
- data/spec/weka/classifiers/trees/balanced_random_forest_spec.rb +37 -0
- data/spec/weka/filters/supervised/instance/smote_spec.rb +5 -0
- data/wikipedia-vandalism_detection.gemspec +37 -0
- metadata +550 -0
@@ -0,0 +1,94 @@
|
|
1
|
+
require 'active_support/core_ext/string'
|
2
|
+
require 'active_support/core_ext/array'
|
3
|
+
|
4
|
+
require 'wikipedia/vandalism_detection/wikitext_extractor'
|
5
|
+
require 'wikipedia/vandalism_detection/features'
|
6
|
+
require 'wikipedia/vandalism_detection/edit'
|
7
|
+
|
8
|
+
module Wikipedia
|
9
|
+
module VandalismDetection
|
10
|
+
# This class provides methods for calculating a feature set of an edit.
|
11
|
+
# The features that shall be used can be defined in the config/wikipedia-vandalism-detection.yml file
|
12
|
+
# under the 'features:' root attribute like this:
|
13
|
+
#
|
14
|
+
# features:
|
15
|
+
# - anonymity
|
16
|
+
# - character sequence
|
17
|
+
# - ...
|
18
|
+
# etc.
|
19
|
+
class FeatureCalculator
|
20
|
+
def initialize
|
21
|
+
@features = Wikipedia::VandalismDetection.config.features
|
22
|
+
raise FeaturesNotConfiguredError if @features.blank? || @features.empty?
|
23
|
+
@feature_classes = build_feature_classes @features
|
24
|
+
end
|
25
|
+
|
26
|
+
# Calculates the configured festures for the given edit and returns an
|
27
|
+
# array of the computed values.
|
28
|
+
def calculate_features_for(edit)
|
29
|
+
raise ArgumentError, 'Input has to be an Edit.' unless edit.is_a?(Edit)
|
30
|
+
|
31
|
+
features = @feature_classes.map do |feature|
|
32
|
+
begin
|
33
|
+
feature.calculate(edit)
|
34
|
+
rescue WikitextExtractionError
|
35
|
+
$stderr.print %{
|
36
|
+
Edit (#{edit.old_revision.id}, #{edit.new_revision.id}) could not
|
37
|
+
be parsed by the WikitextExtractor and will be discarded.\n""}
|
38
|
+
|
39
|
+
Features::MISSING_VALUE
|
40
|
+
end
|
41
|
+
end
|
42
|
+
|
43
|
+
features
|
44
|
+
end
|
45
|
+
|
46
|
+
# Returns the calculated Numeric feature value for given edit and feature with given name
|
47
|
+
def calculate_feature_for(edit, feature_name)
|
48
|
+
unless edit.is_a?(Edit)
|
49
|
+
raise ArgumentError, 'First parameter has to be an Edit.'
|
50
|
+
end
|
51
|
+
|
52
|
+
unless feature_name.is_a?(String)
|
53
|
+
message = 'Second parameter has to be a feature name String ' \
|
54
|
+
'(e.g. "anonymity").'
|
55
|
+
raise ArgumentError, message
|
56
|
+
end
|
57
|
+
|
58
|
+
value = Features::MISSING_VALUE
|
59
|
+
|
60
|
+
begin
|
61
|
+
feature = feature_class_from_name(feature_name)
|
62
|
+
value = feature.calculate(edit)
|
63
|
+
rescue WikitextExtractionError
|
64
|
+
$stderr.print %{
|
65
|
+
Edit (#{edit.old_revision.id}, #{edit.new_revision.id}) could not
|
66
|
+
be parsed by the WikitextExtractor and will be discarded.\n""}
|
67
|
+
end
|
68
|
+
|
69
|
+
value
|
70
|
+
end
|
71
|
+
|
72
|
+
# Returns the feature names as defined in
|
73
|
+
# conf/wikipedia-vandalism-detection.yml under 'features:'.
|
74
|
+
def used_features
|
75
|
+
@features
|
76
|
+
end
|
77
|
+
|
78
|
+
private
|
79
|
+
|
80
|
+
# Returns an array of all configured Feature class instances.
|
81
|
+
def build_feature_classes(feature_names)
|
82
|
+
feature_names.map do |name|
|
83
|
+
feature_class_from_name(name)
|
84
|
+
end
|
85
|
+
end
|
86
|
+
|
87
|
+
# Returns the Feature class of the given name
|
88
|
+
def feature_class_from_name(name)
|
89
|
+
camelcased_name = name.split(/[\s-]/).map(&:capitalize!).join('')
|
90
|
+
"Wikipedia::VandalismDetection::Features::#{camelcased_name}".constantize.new
|
91
|
+
end
|
92
|
+
end
|
93
|
+
end
|
94
|
+
end
|
@@ -0,0 +1,22 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/frequency_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists'
|
3
|
+
require 'wikipedia/vandalism_detection/text'
|
4
|
+
|
5
|
+
module Wikipedia
|
6
|
+
module VandalismDetection
|
7
|
+
module Features
|
8
|
+
# This feature computes frequency of all wordlists words in the inserted
|
9
|
+
# text.
|
10
|
+
class AllWordlistsFrequency < FrequencyBase
|
11
|
+
# Returns the percentage of wordlists words in the inserted text.
|
12
|
+
# Returns 0.0 if inserted clean text is of zero length.
|
13
|
+
def calculate(edit)
|
14
|
+
super
|
15
|
+
|
16
|
+
text = Text.new(edit.inserted_words.join("\n")).clean
|
17
|
+
frequency(text, WordLists.all)
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
22
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/impact_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
# This feature computes the percentage by which the edit increases the
|
8
|
+
# number of all wordlists words in the text.
|
9
|
+
class AllWordlistsImpact < ImpactBase
|
10
|
+
def calculate(edit)
|
11
|
+
super
|
12
|
+
|
13
|
+
old_text = edit.old_revision.text.clean
|
14
|
+
new_text = edit.new_revision.text.clean
|
15
|
+
|
16
|
+
impact(old_text, new_text, WordLists.all)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,17 @@
|
|
1
|
+
require_relative 'base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
# This feature describes whether the contributor of the new revision is
|
7
|
+
# an anonymous or registered Wikipedia user.
|
8
|
+
class Anonymity < Base
|
9
|
+
def calculate(edit)
|
10
|
+
super
|
11
|
+
|
12
|
+
edit.new_revision.anonymous_contributor? ? 0 : 1
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|
@@ -0,0 +1,32 @@
|
|
1
|
+
require_relative 'base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
# This feature describes whether the contributor of the old revision is
|
7
|
+
# an anonymous or registered Wikipedia user.
|
8
|
+
class AnonymityPrevious < Base
|
9
|
+
def calculate(edit)
|
10
|
+
super
|
11
|
+
|
12
|
+
old_revision = edit.old_revision
|
13
|
+
|
14
|
+
if old_revision.contributor.blank?
|
15
|
+
xml = Wikipedia.api_request(
|
16
|
+
prop: 'revisions',
|
17
|
+
rvprop: 'user',
|
18
|
+
revids: old_revision.id
|
19
|
+
)
|
20
|
+
|
21
|
+
contributor = xml.xpath('//rev/@user').text
|
22
|
+
return Features::MISSING_VALUE if contributor.blank?
|
23
|
+
|
24
|
+
old_revision.contributor = contributor
|
25
|
+
end
|
26
|
+
|
27
|
+
old_revision.anonymous_contributor? ? 0 : 1
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
31
|
+
end
|
32
|
+
end
|
@@ -0,0 +1,17 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
# This feature computes the size of the edit's new revision text
|
7
|
+
# (article size).
|
8
|
+
class ArticleSize < Base
|
9
|
+
def calculate(edit)
|
10
|
+
super
|
11
|
+
|
12
|
+
edit.new_revision.text.size
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/frequency_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/bad'
|
3
|
+
require 'wikipedia/vandalism_detection/text'
|
4
|
+
|
5
|
+
module Wikipedia
|
6
|
+
module VandalismDetection
|
7
|
+
module Features
|
8
|
+
# This feature computes frequency of bad words in the inserted text.
|
9
|
+
class BadFrequency < FrequencyBase
|
10
|
+
# Returns the percentage of bad words in the inserted text.
|
11
|
+
# Returns 0.0 if inserted clean text is of zero length.
|
12
|
+
def calculate(edit)
|
13
|
+
super
|
14
|
+
|
15
|
+
text = Text.new(edit.inserted_words.join("\n")).clean
|
16
|
+
frequency(text, WordLists::BAD)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,20 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/impact_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/bad'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
# This feature computes the percentage by which the edit increases the
|
8
|
+
# number of bad words in the text.
|
9
|
+
class BadImpact < ImpactBase
|
10
|
+
def calculate(edit)
|
11
|
+
super
|
12
|
+
old_text = edit.old_revision.text.clean
|
13
|
+
new_text = edit.new_revision.text.clean
|
14
|
+
|
15
|
+
impact(old_text, new_text, WordLists::BAD)
|
16
|
+
end
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
@@ -0,0 +1,61 @@
|
|
1
|
+
require 'wikipedia'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
MISSING_VALUE = '?'.freeze
|
7
|
+
|
8
|
+
# This class should be the base class for all Wikipedia::Feature classes.
|
9
|
+
class Base
|
10
|
+
# Base method for feature calculation.
|
11
|
+
# This method should be overwritten in the concrete
|
12
|
+
# Wikipedia::Feature-classes.
|
13
|
+
#
|
14
|
+
# @example
|
15
|
+
# def calculate(edit)
|
16
|
+
# super # to handle ArgumentException
|
17
|
+
#
|
18
|
+
# ... concrete calculation of feature out of edit...
|
19
|
+
# end
|
20
|
+
def calculate(edit)
|
21
|
+
return if edit.is_a?(Edit)
|
22
|
+
raise ArgumentError, 'Passed argument has to be an Edit'
|
23
|
+
end
|
24
|
+
|
25
|
+
# Count the apperance of a given single term or multiple terms in the
|
26
|
+
# given text
|
27
|
+
#
|
28
|
+
# @param terms String
|
29
|
+
# @param options Hash of form { in: String }
|
30
|
+
#
|
31
|
+
# @example
|
32
|
+
# feature.count "and", in: text
|
33
|
+
# feature.count ["and", "or"], in: text
|
34
|
+
#
|
35
|
+
# @return Integer
|
36
|
+
def count(terms, options = {})
|
37
|
+
unless options[:in]
|
38
|
+
raise ArgumentError, 'The options hash must include the in: key'
|
39
|
+
end
|
40
|
+
|
41
|
+
unless terms.is_a?(String) || terms.is_a?(Array)
|
42
|
+
raise ArgumentError, 'The 1st arg should be an Array or String'
|
43
|
+
end
|
44
|
+
|
45
|
+
words = options[:in].downcase
|
46
|
+
freq = Hash.new(0)
|
47
|
+
|
48
|
+
words.gsub(/[\.,'{2,}:\!\?\(\)]/, '').split.each do |word|
|
49
|
+
freq[word.to_sym] += 1
|
50
|
+
end
|
51
|
+
|
52
|
+
if terms.is_a?(String)
|
53
|
+
freq[terms.downcase.to_sym]
|
54
|
+
else
|
55
|
+
terms.reduce(0) { |result, term| result + freq[term] }
|
56
|
+
end
|
57
|
+
end
|
58
|
+
end
|
59
|
+
end
|
60
|
+
end
|
61
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/frequency_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/biased'
|
3
|
+
require 'wikipedia/vandalism_detection/text'
|
4
|
+
|
5
|
+
module Wikipedia
|
6
|
+
module VandalismDetection
|
7
|
+
module Features
|
8
|
+
# This feature computes frequency of biased words in the inserted text.
|
9
|
+
class BiasedFrequency < FrequencyBase
|
10
|
+
# Returns the percentage of biased words in the inserted text.
|
11
|
+
# Returns 0.0 if inserted clean text is of zero length.
|
12
|
+
def calculate(edit)
|
13
|
+
super
|
14
|
+
|
15
|
+
text = Text.new(edit.inserted_words.join("\n")).clean
|
16
|
+
frequency(text, WordLists::BIASED)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/impact_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/biased'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
# This feature computes the percentage by which the edit increases the
|
8
|
+
# number of biased words in the text.
|
9
|
+
class BiasedImpact < ImpactBase
|
10
|
+
def calculate(edit)
|
11
|
+
super
|
12
|
+
|
13
|
+
old_text = edit.old_revision.text.clean
|
14
|
+
new_text = edit.new_revision.text.clean
|
15
|
+
|
16
|
+
impact(old_text, new_text, WordLists::BIASED)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,26 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
# This feature returns whether the edit's is a blanking.
|
7
|
+
# size < 7, based on Mola Velasco 2010 implementation.
|
8
|
+
class Blanking < Base
|
9
|
+
BLANKING_THRESHOLD = 7
|
10
|
+
|
11
|
+
def calculate(edit)
|
12
|
+
super
|
13
|
+
|
14
|
+
old_text_size = edit.old_revision.text.size
|
15
|
+
new_text_size = edit.new_revision.text.size
|
16
|
+
|
17
|
+
text_removed = old_text_size > new_text_size
|
18
|
+
above_threshold = new_text_size < BLANKING_THRESHOLD
|
19
|
+
|
20
|
+
blanking = text_removed && above_threshold
|
21
|
+
blanking ? 1.0 : 0.0
|
22
|
+
end
|
23
|
+
end
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
@@ -0,0 +1,23 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
# This feature computes the character diversity of the edit's new revision inserted text.
|
7
|
+
# I.e. how many unique characters are amongst all inserted?
|
8
|
+
#
|
9
|
+
# Random typing leads to less unique characters relative to full length =>
|
10
|
+
class CharacterDiversity < Base
|
11
|
+
def calculate(edit)
|
12
|
+
super
|
13
|
+
|
14
|
+
inserted_letters = edit.inserted_text.scan(/[^\s]/)
|
15
|
+
all_letters_count = inserted_letters.size
|
16
|
+
unique_count = inserted_letters.uniq.size
|
17
|
+
|
18
|
+
all_letters_count**(1.0 / unique_count)
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
22
|
+
end
|
23
|
+
end
|
@@ -0,0 +1,19 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
# This feature computes the length of the longest sequence of the same
|
7
|
+
# character in the inserted text.
|
8
|
+
class CharacterSequence < Base
|
9
|
+
def calculate(edit)
|
10
|
+
super
|
11
|
+
|
12
|
+
sequence_hash = edit.inserted_text.scan(/((.)\2*)/)
|
13
|
+
sequence_hash = sequence_hash.group_by { |seq, _| seq.length }
|
14
|
+
sequence_hash.empty? ? 0 : sequence_hash.max.first
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/frequency_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/bad'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
# This feature computes frequency of bad words in the comment of the
|
8
|
+
# edit's new revision.
|
9
|
+
class CommentBadFrequency < FrequencyBase
|
10
|
+
# Returns the percentage of bad words in the new revision's comment.
|
11
|
+
# Returns 0.0 if text is of zero length.
|
12
|
+
def calculate(edit)
|
13
|
+
super
|
14
|
+
|
15
|
+
comment = edit.new_revision.comment.clean
|
16
|
+
frequency(comment, WordLists::BAD)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/frequency_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/biased'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
# This feature computes frequency of biased words in the comment of the
|
8
|
+
# edit's new revision.
|
9
|
+
class CommentBiasedFrequency < FrequencyBase
|
10
|
+
# Returns the percentage of biased words in the new revision's comment.
|
11
|
+
# Returns 0.0 if text is of zero length.
|
12
|
+
def calculate(edit)
|
13
|
+
super
|
14
|
+
|
15
|
+
comment = edit.new_revision.comment.clean
|
16
|
+
frequency(comment, WordLists::BIASED)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
# This feature computes the byte length of the edit's new revision's
|
7
|
+
# comment.
|
8
|
+
class CommentLength < Base
|
9
|
+
def calculate(edit)
|
10
|
+
super
|
11
|
+
edit.new_revision.comment.clean.bytesize
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,30 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/markup'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
# This feature computes frequency of markup words in the comment of the
|
8
|
+
# edit's new revision.
|
9
|
+
class CommentMarkupFrequency < Base
|
10
|
+
MARKUP_REGEX = /(#{WordLists::MARKUP.join('|')})/
|
11
|
+
|
12
|
+
# Returns the percentage of markup words in the new revision's comment.
|
13
|
+
# Returns 0.0 if text is of zero length.
|
14
|
+
def calculate(edit)
|
15
|
+
super
|
16
|
+
|
17
|
+
comment = edit.new_revision.comment
|
18
|
+
all_words_count = comment.split.count
|
19
|
+
markup_words_count = comment.scan(MARKUP_REGEX).count
|
20
|
+
|
21
|
+
if all_words_count > 0
|
22
|
+
markup_words_count.to_f / all_words_count.to_f
|
23
|
+
else
|
24
|
+
0.0
|
25
|
+
end
|
26
|
+
end
|
27
|
+
end
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/frequency_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/pronouns'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
# This feature computes the frequency of pronouns in the comment of the
|
8
|
+
# new revision.
|
9
|
+
class CommentPronounFrequency < FrequencyBase
|
10
|
+
# Returns the percentage of pronoun words in the new revision's
|
11
|
+
# comment. Returns 0.0 if text is of zero length.
|
12
|
+
def calculate(edit)
|
13
|
+
super
|
14
|
+
|
15
|
+
comment = edit.new_revision.comment.clean
|
16
|
+
frequency(comment, WordLists::PRONOUNS)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/frequency_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/sex'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
# This feature computes frequency of sex words in the comment of the
|
8
|
+
# edit's new revision.
|
9
|
+
class CommentSexFrequency < FrequencyBase
|
10
|
+
# Returns the percentage of sex words in the new revision's comment.
|
11
|
+
# Returns 0.0 if text is of zero length.
|
12
|
+
def calculate(edit)
|
13
|
+
super
|
14
|
+
|
15
|
+
comment = edit.new_revision.comment.clean
|
16
|
+
frequency(comment, WordLists::SEX)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/frequency_base'
|
2
|
+
require 'wikipedia/vandalism_detection/word_lists/vulgarism'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
module Features
|
7
|
+
# This feature computes frequency of vulgarism words in the comment of the
|
8
|
+
# edit's new revision.
|
9
|
+
class CommentVulgarismFrequency < FrequencyBase
|
10
|
+
# Returns the percentage of vulgarism words in the new revision's
|
11
|
+
# comment.Returns 0.0 if text is of zero length.
|
12
|
+
def calculate(edit)
|
13
|
+
super
|
14
|
+
|
15
|
+
comment = edit.new_revision.comment.clean
|
16
|
+
frequency(comment, WordLists::VULGARISM)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,31 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
require 'zlib'
|
3
|
+
require 'wikipedia/vandalism_detection/diff'
|
4
|
+
|
5
|
+
module Wikipedia
|
6
|
+
module VandalismDetection
|
7
|
+
module Features
|
8
|
+
# This feature describes compressibility ratio of compressed and
|
9
|
+
# uncompressed inserted text.
|
10
|
+
class Compressibility < Base
|
11
|
+
# Calculates the compressibility ratio of the inserted text.
|
12
|
+
# Values above 0.5 are higher compressed and therefor can stand for
|
13
|
+
# nonsense text as:
|
14
|
+
# 'AAAAAAAAAAAAAAAAAAAhhhhhhhhhhhhhhhh!' etc.
|
15
|
+
def calculate(edit)
|
16
|
+
super
|
17
|
+
|
18
|
+
inserted_text = edit.inserted_text
|
19
|
+
uncompressed_size = inserted_text.bytesize.to_f
|
20
|
+
compressed_size = Zlib::Deflate.deflate(inserted_text).bytesize.to_f
|
21
|
+
|
22
|
+
if inserted_text.empty?
|
23
|
+
0.5
|
24
|
+
else
|
25
|
+
uncompressed_size / (compressed_size + uncompressed_size)
|
26
|
+
end
|
27
|
+
end
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
31
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
class ContainsBase < Base
|
7
|
+
# Returns whether the comment contains the given term.
|
8
|
+
# It returns 0 if term is not included, else 1.
|
9
|
+
def contains(comment, terms)
|
10
|
+
terms = terms.is_a?(Array) ? terms.join('|') : terms
|
11
|
+
comment =~ /#{terms}/i ? 1 : 0
|
12
|
+
end
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,17 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/features/contains_base'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Features
|
6
|
+
# This feature returns whether the edit's comment includes reverted key
|
7
|
+
# words.
|
8
|
+
class Reverted < ContainsBase
|
9
|
+
def calculate(edit)
|
10
|
+
super
|
11
|
+
|
12
|
+
contains(edit.new_revision.comment, %w[rvt rvv revert])
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|