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 +1,12 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/algorithms/kullback_leibler_divergence'
|
2
|
+
|
3
|
+
module Wikipedia
|
4
|
+
module VandalismDetection
|
5
|
+
module Algorithms
|
6
|
+
def kullback_leibler_divergence(text_a, text_b)
|
7
|
+
divergence = KullbackLeiblerDivergence.new
|
8
|
+
divergence.of(text_a, text_b)
|
9
|
+
end
|
10
|
+
end
|
11
|
+
end
|
12
|
+
end
|
@@ -0,0 +1,202 @@
|
|
1
|
+
require 'weka'
|
2
|
+
require 'active_support/core_ext/string'
|
3
|
+
require 'fileutils'
|
4
|
+
|
5
|
+
require 'wikipedia/vandalism_detection/configuration'
|
6
|
+
require 'wikipedia/vandalism_detection/edit'
|
7
|
+
require 'wikipedia/vandalism_detection/feature_calculator'
|
8
|
+
require 'wikipedia/vandalism_detection/instances'
|
9
|
+
require 'wikipedia/vandalism_detection/evaluator'
|
10
|
+
|
11
|
+
module Wikipedia
|
12
|
+
module VandalismDetection
|
13
|
+
class Classifier
|
14
|
+
attr_reader :evaluator, :dataset
|
15
|
+
|
16
|
+
# Loads the classifier instance configured in the config file.
|
17
|
+
def initialize(dataset = nil)
|
18
|
+
@config = Wikipedia::VandalismDetection.config
|
19
|
+
@feature_calculator = FeatureCalculator.new
|
20
|
+
@classifier = load_classifier(dataset)
|
21
|
+
@evaluator = Evaluator.new(self)
|
22
|
+
end
|
23
|
+
|
24
|
+
# Returns the concrete classifier instance configured in the config file
|
25
|
+
# When you configured a Trees::RandomForest classifier you will get a
|
26
|
+
# Weka::Classifiers::Trees::RandomForest instance.
|
27
|
+
# This instance can be used for native function callings of the classifier
|
28
|
+
# class.
|
29
|
+
def classifier_instance
|
30
|
+
@classifier
|
31
|
+
end
|
32
|
+
|
33
|
+
# Classifies an edit or a set of features and returns the vandalism
|
34
|
+
# confidence by default.
|
35
|
+
# If option 'return_all_params: true' is set, it returns a Hash of form
|
36
|
+
# { confidence => ..., class_index => ...}
|
37
|
+
#
|
38
|
+
# @example
|
39
|
+
# # suppose you have a dataset with 2 feature or 'edit' as an instance
|
40
|
+
# # of Wikipedia::VandalismDetection::Edit
|
41
|
+
# classifier = Wikipedia::VandalsimDetection::Classifier.new
|
42
|
+
# features = [0.45, 0.67]
|
43
|
+
#
|
44
|
+
# confidence = classifier.classify(features)
|
45
|
+
# confidence = classifier.classify(edit)
|
46
|
+
def classify(edit_or_features, options = {})
|
47
|
+
features = @config.features
|
48
|
+
param_is_features = edit_or_features.is_a?(Array) && edit_or_features.size == features.count
|
49
|
+
param_is_edit = edit_or_features.is_a? Edit
|
50
|
+
|
51
|
+
unless param_is_edit || param_is_features
|
52
|
+
message = 'Input has to be an Edit or an Array of feature values.'
|
53
|
+
raise ArgumentError, message
|
54
|
+
end
|
55
|
+
|
56
|
+
feature_values = param_is_edit ? @feature_calculator.calculate_features_for(edit_or_features) : edit_or_features
|
57
|
+
return -1.0 if feature_values.empty?
|
58
|
+
|
59
|
+
feature_values = feature_values.map do |i|
|
60
|
+
i == Features::MISSING_VALUE ? nil : i
|
61
|
+
end
|
62
|
+
|
63
|
+
dataset = Instances.empty
|
64
|
+
dataset.set_class_index(feature_values.count)
|
65
|
+
dataset.add_instance([*feature_values, Instances::VANDALISM])
|
66
|
+
|
67
|
+
instance = dataset.instance(0)
|
68
|
+
instance.set_class_missing
|
69
|
+
|
70
|
+
if @config.use_occ?
|
71
|
+
if @config.classifier_options =~ /#{Instances::VANDALISM}/
|
72
|
+
index = Instances::VANDALISM_CLASS_INDEX
|
73
|
+
else
|
74
|
+
index = Instances::REGULAR_CLASS_INDEX
|
75
|
+
end
|
76
|
+
else
|
77
|
+
index = Instances::VANDALISM_CLASS_INDEX
|
78
|
+
end
|
79
|
+
|
80
|
+
|
81
|
+
confidence = @classifier.distribution_for_instance(instance).to_a[index]
|
82
|
+
|
83
|
+
if options[:return_all_params]
|
84
|
+
class_index = @classifier.classify_instance(instance)
|
85
|
+
class_index = class_index.nan? ? Instances::NOT_KNOWN_INDEX : class_index.to_i
|
86
|
+
results = { confidence: confidence, class_index: class_index }
|
87
|
+
else
|
88
|
+
results = confidence
|
89
|
+
end
|
90
|
+
|
91
|
+
results
|
92
|
+
end
|
93
|
+
|
94
|
+
# Cross validates the classifier.
|
95
|
+
# Fold is used as defined in configuration (default is 10).
|
96
|
+
#
|
97
|
+
# @example
|
98
|
+
# classifier = Wikipedia::VandalismDetection::Classifier.new
|
99
|
+
# evaluation = classifier.cross_validate
|
100
|
+
# evaluation = classifier.cross_validate(equally_distributed: true)
|
101
|
+
#
|
102
|
+
def cross_validate(options = {})
|
103
|
+
@evaluator.cross_validate(options)
|
104
|
+
end
|
105
|
+
|
106
|
+
private
|
107
|
+
|
108
|
+
# Loads the (Weka-) Classifier set in the Configuration
|
109
|
+
def load_classifier(dataset)
|
110
|
+
classifier_name = @config.classifier_type
|
111
|
+
|
112
|
+
unless classifier_name
|
113
|
+
message = 'Classifier type is not defined in wikipedia-vandalism-detection.yml'
|
114
|
+
raise ClassifierNotConfiguredError, message
|
115
|
+
end
|
116
|
+
|
117
|
+
if @config.features.blank?
|
118
|
+
message = 'No features configured in wikipedia-vandalism-detection.yml'
|
119
|
+
raise FeaturesNotConfiguredError, message
|
120
|
+
end
|
121
|
+
|
122
|
+
begin
|
123
|
+
"Weka::Classifiers::#{classifier_name}".constantize
|
124
|
+
rescue
|
125
|
+
message = "The configured classifier type '#{classifier_name}' is unknown."
|
126
|
+
raise ClassifierUnknownError, message
|
127
|
+
end
|
128
|
+
|
129
|
+
classifier_class = "Weka::Classifiers::#{classifier_name}".constantize
|
130
|
+
options = @config.classifier_options
|
131
|
+
|
132
|
+
puts "Loading classifier #{classifier_name} with options '#{options}'…"
|
133
|
+
|
134
|
+
if dataset.nil?
|
135
|
+
if @config.balanced_training_data?
|
136
|
+
puts 'using BALANCED training dataset'
|
137
|
+
dataset = TrainingDataset.balanced_instances
|
138
|
+
elsif @config.unbalanced_training_data?
|
139
|
+
puts 'using FULL (unbalanced) training dataset'
|
140
|
+
dataset = TrainingDataset.instances
|
141
|
+
elsif @config.oversampled_training_data?
|
142
|
+
puts 'using OVERSAMPLED training dataset'
|
143
|
+
dataset = TrainingDataset.oversampled_instances
|
144
|
+
end
|
145
|
+
end
|
146
|
+
|
147
|
+
if @config.use_occ?
|
148
|
+
dataset.rename_attribute_value(
|
149
|
+
dataset.class_index,
|
150
|
+
one_class_index,
|
151
|
+
Instances::OUTLIER
|
152
|
+
)
|
153
|
+
end
|
154
|
+
|
155
|
+
@dataset = dataset
|
156
|
+
|
157
|
+
begin
|
158
|
+
classifier = classifier_class.build do
|
159
|
+
use_options options if options
|
160
|
+
train_with_instances dataset
|
161
|
+
end
|
162
|
+
|
163
|
+
classifier
|
164
|
+
rescue => error
|
165
|
+
raise "Error while loading classifier: #{error}"
|
166
|
+
end
|
167
|
+
end
|
168
|
+
|
169
|
+
def one_class_index
|
170
|
+
if @config.classifier_options =~ /#{Instances::VANDALISM}/
|
171
|
+
Instances::REGULAR_CLASS_INDEX
|
172
|
+
else
|
173
|
+
Instances::VANDALISM_CLASS_INDEX
|
174
|
+
end
|
175
|
+
end
|
176
|
+
|
177
|
+
# Returns the given dataset cleaned up the regular instances
|
178
|
+
def remove_regular_instances(dataset)
|
179
|
+
features = @config.features
|
180
|
+
|
181
|
+
vandalism_dataset = Weka::Core::Instances.new.with_attributes do
|
182
|
+
features.each { |name| numeric :"#{name.tr(' ', '_')}" }
|
183
|
+
nominal :class, values: [Instances::VANDALISM], class_attribute: true
|
184
|
+
end
|
185
|
+
|
186
|
+
dataset.to_a.map(&:values).each_with_index do |attributes, index|
|
187
|
+
class_value = Instances::CLASSES[dataset.instance(index).value(dataset.class_index).to_i]
|
188
|
+
|
189
|
+
if class_value == Instances::VANDALISM
|
190
|
+
values = attributes[0..-2]
|
191
|
+
vandalism_dataset.add_instance([*values, class_value])
|
192
|
+
end
|
193
|
+
end
|
194
|
+
|
195
|
+
filter = Weka::Filters::Unsupervised::Attribute::Normalize.new
|
196
|
+
vandalism_dataset = filter.filter(vandalism_dataset)
|
197
|
+
|
198
|
+
vandalism_dataset
|
199
|
+
end
|
200
|
+
end
|
201
|
+
end
|
202
|
+
end
|
@@ -0,0 +1,350 @@
|
|
1
|
+
require 'weka/classifiers/meta/one_class_classifier'
|
2
|
+
require 'singleton'
|
3
|
+
|
4
|
+
module Wikipedia
|
5
|
+
module VandalismDetection
|
6
|
+
require 'yaml'
|
7
|
+
|
8
|
+
def self.config
|
9
|
+
Configuration.instance
|
10
|
+
end
|
11
|
+
|
12
|
+
class Configuration
|
13
|
+
include Singleton
|
14
|
+
|
15
|
+
TRAINING_DATA_BALANCED = 'balanced'.freeze
|
16
|
+
TRAINING_DATA_UNBALANCED = 'unbalanced'.freeze
|
17
|
+
TRAINING_DATA_OVERSAMPLED = 'oversampled'.freeze
|
18
|
+
CONFIG_FILE = 'wikipedia-vandalism-detection.yml'.freeze
|
19
|
+
|
20
|
+
attr_reader :data,
|
21
|
+
:features,
|
22
|
+
:classifier_options,
|
23
|
+
:classifier_type,
|
24
|
+
:cross_validation_fold,
|
25
|
+
:output_base_directory,
|
26
|
+
:training_data_options
|
27
|
+
|
28
|
+
def initialize
|
29
|
+
config = DefaultConfiguration[DefaultConfiguration::DEFAULTS]
|
30
|
+
@config_from_file ||= config.load_config_file(config.config_file)
|
31
|
+
|
32
|
+
@data ||= @config_from_file ? config.deep_merge(@config_from_file) : config
|
33
|
+
|
34
|
+
@classifier_type = @data['classifier']['type']
|
35
|
+
@classifier_options = @data['classifier']['options']
|
36
|
+
@cross_validation_fold = @data['classifier']['cross-validation-fold']
|
37
|
+
@training_data_options = @data['classifier']['training-data-options']
|
38
|
+
@replace_missing_values = @data['classifier']['replace-missing-values'].to_s
|
39
|
+
|
40
|
+
@features = @data['features']
|
41
|
+
@output_base_directory = File.expand_path(@data['output']['base_directory'], __FILE__)
|
42
|
+
@training_arff_file_name = @data['output']['training']['arff_file']
|
43
|
+
@test_arff_file_name = @data['output']['test']['arff_file']
|
44
|
+
end
|
45
|
+
|
46
|
+
# Returns whether the classifier uses one class classification
|
47
|
+
def use_occ?
|
48
|
+
@classifier_type == Weka::Classifiers::Meta::OneClassClassifier.type
|
49
|
+
end
|
50
|
+
|
51
|
+
def replace_training_data_missing_values?
|
52
|
+
!!(@replace_missing_values =~ /(true|t|yes|y)/i)
|
53
|
+
end
|
54
|
+
|
55
|
+
# Returns a boolean value whether a balanced data set is used for
|
56
|
+
# classifier training.
|
57
|
+
# (balanced means: same number of vandalism and regular samples)
|
58
|
+
def balanced_training_data?
|
59
|
+
@training_data_options == TRAINING_DATA_BALANCED
|
60
|
+
end
|
61
|
+
|
62
|
+
# Returns a boolean value whether an unbalanced data set is used for
|
63
|
+
# classifier training.
|
64
|
+
# (unbalanced means: vandalism and regular samples are used as given in
|
65
|
+
# arff file)
|
66
|
+
def unbalanced_training_data?
|
67
|
+
@training_data_options == TRAINING_DATA_UNBALANCED ||
|
68
|
+
@training_data_options.nil? ||
|
69
|
+
(!balanced_training_data? && !oversampled_training_data?)
|
70
|
+
end
|
71
|
+
|
72
|
+
# Returns a boolean value whether a oversampled data set is used for
|
73
|
+
# classifier training.
|
74
|
+
# (oversampled means: a balanced dataset is enriched through vandalism
|
75
|
+
# instances if vandalism number is less than regular number)
|
76
|
+
def oversampled_training_data?
|
77
|
+
!@training_data_options.nil? &&
|
78
|
+
@training_data_options.include?(TRAINING_DATA_OVERSAMPLED)
|
79
|
+
end
|
80
|
+
|
81
|
+
# Returns a hash of the oversampled training data options.
|
82
|
+
# Allowed options are -p (-percent) and -u (-undersampling)
|
83
|
+
def oversampling_options
|
84
|
+
if oversampled_training_data?
|
85
|
+
params = @training_data_options.gsub(TRAINING_DATA_OVERSAMPLED, '').split('-')
|
86
|
+
|
87
|
+
percent_default = 100.0
|
88
|
+
undersampling_default = 100.0
|
89
|
+
|
90
|
+
percent_option = params.select { |param| param.match(/(p\s|percentage\s)\d+/i) }[0]
|
91
|
+
undersampling_option = params.select { |param| param.match(/(u\s|undersampling\s)/i) }[0]
|
92
|
+
|
93
|
+
percent = percent_option.nil? ? percent_default : percent_option.split.last.to_f
|
94
|
+
undersampling = undersampling_default
|
95
|
+
|
96
|
+
if undersampling_option
|
97
|
+
if !undersampling_option.match(/(true|t|yes|y)/i).nil?
|
98
|
+
undersampling_percentage = undersampling_option.split.last
|
99
|
+
undersampling = undersampling_percentage.nil? ? undersampling_default : undersampling_percentage.to_f
|
100
|
+
else
|
101
|
+
undersampling = 0.0
|
102
|
+
end
|
103
|
+
end
|
104
|
+
|
105
|
+
{ percentage: percent, undersampling: undersampling }
|
106
|
+
else
|
107
|
+
{}
|
108
|
+
end
|
109
|
+
end
|
110
|
+
|
111
|
+
# Returns the path to the classification file.
|
112
|
+
# Automatically sub directories for classifier and training data options
|
113
|
+
# are added. Thus it results in
|
114
|
+
# <output base dir>/<classifier name>/<training data options>/<file name>
|
115
|
+
def test_output_classification_file
|
116
|
+
classifiction_file_name = @data['output']['test']['classification_file']
|
117
|
+
classifier_name = @classifier_type.split('::').last.downcase
|
118
|
+
|
119
|
+
File.join(
|
120
|
+
@output_base_directory,
|
121
|
+
classifier_name,
|
122
|
+
@training_data_options.gsub(/\s+/, '_'),
|
123
|
+
classifiction_file_name
|
124
|
+
)
|
125
|
+
end
|
126
|
+
|
127
|
+
# Returns the training arff file name.
|
128
|
+
# The path is expanded by used classifier & options and is in the same
|
129
|
+
# directory as the classification file.
|
130
|
+
def training_output_arff_file
|
131
|
+
directory = File.dirname(test_output_classification_file)
|
132
|
+
File.join(directory, @training_arff_file_name)
|
133
|
+
end
|
134
|
+
|
135
|
+
# Returns the test arff file name.
|
136
|
+
# The path is expanded by used classifier & options and is in the same
|
137
|
+
# directory as the classification file.
|
138
|
+
def test_output_arff_file
|
139
|
+
directory = File.dirname(test_output_classification_file)
|
140
|
+
File.join(directory, @test_arff_file_name)
|
141
|
+
end
|
142
|
+
|
143
|
+
# Returns file/path string for corpora files/directories and output files
|
144
|
+
# after following schema: <corpus type>_<progress stage>_<file name>.
|
145
|
+
#
|
146
|
+
# Instead of 'corpora' the word 'corpus' is used for grammatical reasons.
|
147
|
+
#
|
148
|
+
# example:
|
149
|
+
# training_corpus_edits_file()
|
150
|
+
# test_output_index_file()
|
151
|
+
#
|
152
|
+
def method_missing(method_name, *args)
|
153
|
+
if instance_variable_defined?("@#{method_name}")
|
154
|
+
return instance_variable_get("@#{method_name}")
|
155
|
+
end
|
156
|
+
|
157
|
+
file_path_parts = method_name.to_s.split('_')
|
158
|
+
|
159
|
+
if file_path_parts.count >= 4
|
160
|
+
corpus_type = file_path_parts[0]
|
161
|
+
progress_stage = file_path_parts[1]
|
162
|
+
file_path = file_path_parts[2..-1].join('_')
|
163
|
+
|
164
|
+
if progress_stage == 'corpus'
|
165
|
+
progress_stage = 'corpora'
|
166
|
+
path = File.join(
|
167
|
+
@data[progress_stage]['base_directory'],
|
168
|
+
@data[progress_stage][corpus_type]['base_directory']
|
169
|
+
)
|
170
|
+
elsif progress_stage == 'output'
|
171
|
+
path = @output_base_directory
|
172
|
+
else
|
173
|
+
return super
|
174
|
+
end
|
175
|
+
|
176
|
+
relative_path = File.join(path, @data[progress_stage][corpus_type][file_path])
|
177
|
+
absolute_path = File.expand_path(relative_path, __FILE__)
|
178
|
+
instance_variable_set("@#{method_name}", absolute_path)
|
179
|
+
else
|
180
|
+
super
|
181
|
+
end
|
182
|
+
end
|
183
|
+
end
|
184
|
+
|
185
|
+
# This class represents the default config which is merged with the
|
186
|
+
# customized config from config YAML file.
|
187
|
+
class DefaultConfiguration < Hash
|
188
|
+
DEFAULTS = {
|
189
|
+
'source' => Dir.pwd,
|
190
|
+
'features' => [
|
191
|
+
'anonymity',
|
192
|
+
'anonymity previous',
|
193
|
+
'all wordlists frequency',
|
194
|
+
'all wordlists impact',
|
195
|
+
'article size',
|
196
|
+
'bad frequency',
|
197
|
+
'bad impact',
|
198
|
+
'biased frequency',
|
199
|
+
'biased impact',
|
200
|
+
'blanking',
|
201
|
+
'character sequence',
|
202
|
+
'character diversity',
|
203
|
+
'comment length',
|
204
|
+
'comment biased frequency',
|
205
|
+
'comment pronoun frequency',
|
206
|
+
'comment vulgarism frequency',
|
207
|
+
'compressibility',
|
208
|
+
'copyedit',
|
209
|
+
'digit ratio',
|
210
|
+
'edits per user',
|
211
|
+
'emoticons frequency',
|
212
|
+
'emoticons impact',
|
213
|
+
'inserted size',
|
214
|
+
'inserted words',
|
215
|
+
'inserted character distribution',
|
216
|
+
'inserted external links',
|
217
|
+
'inserted internal links',
|
218
|
+
'longest word',
|
219
|
+
'markup frequency',
|
220
|
+
'markup impact',
|
221
|
+
'non-alphanumeric ratio',
|
222
|
+
'personal life',
|
223
|
+
'pronoun frequency',
|
224
|
+
'pronoun impact',
|
225
|
+
'removed size',
|
226
|
+
'removed words',
|
227
|
+
'removed all wordlists frequency',
|
228
|
+
'removed bad frequency',
|
229
|
+
'removed biased frequency',
|
230
|
+
'removed character distribution',
|
231
|
+
'removed emoticons frequency',
|
232
|
+
'removed markup frequency',
|
233
|
+
'removed pronoun frequency',
|
234
|
+
'removed sex frequency',
|
235
|
+
'removed vulgarism frequency',
|
236
|
+
'replacement similarity',
|
237
|
+
'reverted',
|
238
|
+
'revisions character distribution',
|
239
|
+
'sex frequency',
|
240
|
+
'sex impact',
|
241
|
+
'same editor',
|
242
|
+
'size increment',
|
243
|
+
'size ratio',
|
244
|
+
'term frequency',
|
245
|
+
'time interval',
|
246
|
+
'time of day',
|
247
|
+
'upper case ratio',
|
248
|
+
'upper case words ratio',
|
249
|
+
'upper to lower case ratio',
|
250
|
+
'vulgarism frequency',
|
251
|
+
'vulgarism impact',
|
252
|
+
'weekday',
|
253
|
+
'words increment'
|
254
|
+
],
|
255
|
+
'corpora' => {
|
256
|
+
'base_directory' => nil,
|
257
|
+
'training' => {
|
258
|
+
'base_directory' => nil,
|
259
|
+
'edits_file' => nil,
|
260
|
+
'annotations_file' => nil,
|
261
|
+
'revisions_directory' => nil
|
262
|
+
},
|
263
|
+
'test' => {
|
264
|
+
'base_directory' => nil,
|
265
|
+
'edits_file' => nil,
|
266
|
+
'revisions_directory' => nil,
|
267
|
+
'ground_truth_file' => nil
|
268
|
+
}
|
269
|
+
},
|
270
|
+
'output' => {
|
271
|
+
'base_directory' => File.join(Dir.pwd, 'build'),
|
272
|
+
'training' => {
|
273
|
+
'arff_file' => 'training.arff',
|
274
|
+
'index_file' => 'training_index.yml'
|
275
|
+
},
|
276
|
+
'test' => {
|
277
|
+
'arff_file' => 'test.arff',
|
278
|
+
'index_file' => 'test_index.yml',
|
279
|
+
'classification_file' => 'classification.txt'
|
280
|
+
}
|
281
|
+
},
|
282
|
+
'classifier' => {
|
283
|
+
'type' => nil,
|
284
|
+
'options' => nil,
|
285
|
+
'cross-validation-fold' => 10,
|
286
|
+
'training-data-options' => 'unbalanced',
|
287
|
+
'replace-missing-values' => nil
|
288
|
+
}
|
289
|
+
}.freeze
|
290
|
+
|
291
|
+
def source
|
292
|
+
DEFAULTS['source']
|
293
|
+
end
|
294
|
+
|
295
|
+
# Looks in two places for a custom config file:
|
296
|
+
# in <app_root>/config/ and in <app_root>/lib/config
|
297
|
+
def config_file
|
298
|
+
config_file_path = "config/#{Configuration::CONFIG_FILE}"
|
299
|
+
root_file = File.join(source, config_file_path)
|
300
|
+
lib_file = File.join(source, "lib/#{config_file_path}")
|
301
|
+
|
302
|
+
first_parent_file = find_first_parent_path_for(
|
303
|
+
File.expand_path(File.dirname(__FILE__)),
|
304
|
+
config_file_path
|
305
|
+
)
|
306
|
+
|
307
|
+
if File.exist?(root_file)
|
308
|
+
root_file
|
309
|
+
elsif File.exist?(lib_file)
|
310
|
+
lib_file
|
311
|
+
else
|
312
|
+
first_parent_file
|
313
|
+
end
|
314
|
+
end
|
315
|
+
|
316
|
+
def load_config_file(file)
|
317
|
+
config_file = Configuration::CONFIG_FILE
|
318
|
+
|
319
|
+
if File.exist?(file) && file =~ /#{config_file}/
|
320
|
+
YAML.load_file(file)
|
321
|
+
else
|
322
|
+
warn %(
|
323
|
+
|
324
|
+
Configuration file not found in
|
325
|
+
#{source}/config,
|
326
|
+
#{source}/lib/config directory
|
327
|
+
or any other parent path.
|
328
|
+
|
329
|
+
To customize the system, create a '#{config_file}' file.
|
330
|
+
|
331
|
+
)
|
332
|
+
end
|
333
|
+
end
|
334
|
+
|
335
|
+
private
|
336
|
+
|
337
|
+
def find_first_parent_path_for(start_path, file)
|
338
|
+
file_path = File.join(start_path, file)
|
339
|
+
|
340
|
+
if File.exist?(file_path)
|
341
|
+
file_path
|
342
|
+
elsif File.dirname(start_path) != start_path
|
343
|
+
find_first_parent_path_for(File.dirname(start_path), file)
|
344
|
+
else
|
345
|
+
start_path
|
346
|
+
end
|
347
|
+
end
|
348
|
+
end
|
349
|
+
end
|
350
|
+
end
|
@@ -0,0 +1,36 @@
|
|
1
|
+
module Wikipedia
|
2
|
+
module VandalismDetection
|
3
|
+
require 'java'
|
4
|
+
require 'java/diffutils-1.3.0.jar'
|
5
|
+
|
6
|
+
java_import 'difflib.DiffUtils'
|
7
|
+
|
8
|
+
class Diff
|
9
|
+
def initialize(original, current)
|
10
|
+
@original = clean_text(original)
|
11
|
+
@current = clean_text(current)
|
12
|
+
@patch = DiffUtils.diff(@original.split, @current.split)
|
13
|
+
end
|
14
|
+
|
15
|
+
def inserted_words
|
16
|
+
@patch.deltas.map { |delta| delta.revised.lines }.flatten
|
17
|
+
end
|
18
|
+
|
19
|
+
def removed_words
|
20
|
+
@patch.deltas.map { |delta| delta.original.lines }.flatten
|
21
|
+
end
|
22
|
+
|
23
|
+
private
|
24
|
+
|
25
|
+
def clean_text(text)
|
26
|
+
text.encode(
|
27
|
+
'UTF-8',
|
28
|
+
'binary',
|
29
|
+
invalid: :replace,
|
30
|
+
undef: :replace,
|
31
|
+
replace: ''
|
32
|
+
)
|
33
|
+
end
|
34
|
+
end
|
35
|
+
end
|
36
|
+
end
|
@@ -0,0 +1,81 @@
|
|
1
|
+
require 'wikipedia/vandalism_detection/diff'
|
2
|
+
require 'wikipedia/vandalism_detection/text'
|
3
|
+
require 'wikipedia/vandalism_detection/page'
|
4
|
+
|
5
|
+
module Wikipedia
|
6
|
+
module VandalismDetection
|
7
|
+
class Edit
|
8
|
+
attr_reader :old_revision, :new_revision
|
9
|
+
attr_accessor :page
|
10
|
+
|
11
|
+
def initialize(old_revision, new_revision, attributes = {})
|
12
|
+
message = "old revision: #{old_revision.id} | parent: #{old_revision.parent_id},
|
13
|
+
new revision: #{new_revision.id} | parent: #{new_revision.parent_id}"
|
14
|
+
|
15
|
+
unless sequent?(old_revision, new_revision)
|
16
|
+
raise ArgumentError, "Revisions are not sequent: #{message}."
|
17
|
+
end
|
18
|
+
|
19
|
+
@old_revision = old_revision
|
20
|
+
@new_revision = new_revision
|
21
|
+
@page = attributes[:page] || Page.new
|
22
|
+
end
|
23
|
+
|
24
|
+
def serialize(*attributes)
|
25
|
+
old_revision_parts = []
|
26
|
+
new_revision_parts = []
|
27
|
+
|
28
|
+
attributes.each do |attr|
|
29
|
+
if @old_revision.respond_to?(attr)
|
30
|
+
old_revision_parts.push @old_revision.method(attr).call
|
31
|
+
end
|
32
|
+
end
|
33
|
+
|
34
|
+
attributes.each do |attr|
|
35
|
+
if @new_revision.respond_to?(attr)
|
36
|
+
new_revision_parts.push @new_revision.method(attr).call
|
37
|
+
end
|
38
|
+
end
|
39
|
+
|
40
|
+
old_revision_string = old_revision_parts.join(',')
|
41
|
+
new_revision_string = new_revision_parts.join(',')
|
42
|
+
|
43
|
+
"#{old_revision_string}\t#{new_revision_string}"
|
44
|
+
end
|
45
|
+
|
46
|
+
# Returns an array of the words inserted in the new revision compared with
|
47
|
+
# the old one.
|
48
|
+
def inserted_words
|
49
|
+
@diff ||= Diff.new(@old_revision.text, @new_revision.text)
|
50
|
+
@inserted_words ||= @diff.inserted_words
|
51
|
+
end
|
52
|
+
|
53
|
+
# Returns a Text of the words inserted in the new revision compared with
|
54
|
+
# the old one.
|
55
|
+
def inserted_text
|
56
|
+
@inserted_text ||= Text.new(inserted_words.join(' '))
|
57
|
+
end
|
58
|
+
|
59
|
+
# Returns an array of the words removed in the new revision compared with
|
60
|
+
# the old one.
|
61
|
+
def removed_words
|
62
|
+
@diff ||= Diff.new(@old_revision.text, @new_revision.text)
|
63
|
+
@removed_words ||= @diff.removed_words
|
64
|
+
end
|
65
|
+
|
66
|
+
# Returns a Text of the words removed in the new revision compared with
|
67
|
+
# the old one.
|
68
|
+
def removed_text
|
69
|
+
@removed_text ||= Text.new(removed_words.join(' '))
|
70
|
+
end
|
71
|
+
|
72
|
+
protected
|
73
|
+
|
74
|
+
# Returns whether the given revisions are sequent, i.e. the old revisions
|
75
|
+
# id is the the new revisions parent id.
|
76
|
+
def sequent?(old_revision, new_revision)
|
77
|
+
new_revision.parent_id == old_revision.id
|
78
|
+
end
|
79
|
+
end
|
80
|
+
end
|
81
|
+
end
|