commendo 1.2.4 → 2.0.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +3 -0
- data/bin/commendo-create-mysql-db +3 -0
- data/bin/commendo-create.sql +99 -0
- data/bin/commendo-load-tsv +11 -5
- data/bin/commendo-load-tsv-mysql.rb +43 -0
- data/bin/commendo-time-mysql.rb +31 -0
- data/commendo.gemspec +4 -2
- data/lib/commendo.rb +24 -0
- data/lib/commendo/configuration.rb +25 -0
- data/lib/commendo/content_set.rb +13 -182
- data/lib/commendo/mysql-backed/content_set.rb +152 -0
- data/lib/commendo/mysql-backed/tag_set.rb +81 -0
- data/lib/commendo/mysql-backed/weighted_group.rb +40 -0
- data/lib/commendo/redis-backed/content_set.rb +194 -0
- data/lib/commendo/{pair_comparison.lua → redis-backed/pair_comparison.lua} +0 -0
- data/lib/commendo/{similarity.lua → redis-backed/similarity.lua} +0 -0
- data/lib/commendo/redis-backed/tag_set.rb +54 -0
- data/lib/commendo/redis-backed/weighted_group.rb +54 -0
- data/lib/commendo/tag_set.rb +6 -42
- data/lib/commendo/version.rb +1 -1
- data/lib/commendo/weighted_group.rb +7 -41
- data/lib/mysql2/client.rb +17 -0
- data/model 2.mwb +0 -0
- data/sql_model.mwb +0 -0
- data/test/configuration_test.rb +71 -0
- data/test/mysql_content_set_test.rb +40 -0
- data/test/mysql_tag_set_test.rb +34 -0
- data/test/mysql_weighted_group_test.rb +54 -0
- data/test/redis_content_set_test.rb +57 -0
- data/test/redis_tag_set_test.rb +31 -0
- data/test/redis_weighted_group_test.rb +49 -0
- data/test/tests_for_content_sets.rb +379 -0
- data/test/tests_for_tag_sets.rb +130 -0
- data/test/tests_for_weighted_groups.rb +106 -0
- metadata +72 -12
- data/test/content_set_test.rb +0 -408
- data/test/tag_set_test.rb +0 -128
- data/test/weighted_group_test.rb +0 -191
@@ -0,0 +1,130 @@
|
|
1
|
+
module TestsForTagSets
|
2
|
+
|
3
|
+
def test_adds_tags_for_resource
|
4
|
+
assert_equal [], @ts.get(1)
|
5
|
+
@ts.add(1, 'foo', 'bar', 'baz')
|
6
|
+
expected = @ts.get(1)
|
7
|
+
assert_equal %w(bar baz foo).sort, expected.sort
|
8
|
+
@ts.add(1, 'qux', 'qip')
|
9
|
+
expected = @ts.get(1)
|
10
|
+
assert_equal %w(bar baz foo qip qux).sort, expected.sort
|
11
|
+
end
|
12
|
+
|
13
|
+
def test_isolates_by_key_base
|
14
|
+
ts1 = create_tag_set('TagSetOne')
|
15
|
+
ts2 = create_tag_set('TagSetTwo')
|
16
|
+
assert_equal [], ts1.get(1)
|
17
|
+
assert_equal [], ts2.get(1)
|
18
|
+
|
19
|
+
ts1.add(1, 'foo', 'bar', 'baz')
|
20
|
+
|
21
|
+
assert_equal %w(bar baz foo).sort, ts1.get(1).sort
|
22
|
+
assert_equal [], ts2.get(1)
|
23
|
+
|
24
|
+
ts2.add(1, 'qux', 'qip')
|
25
|
+
assert_equal %w(bar baz foo).sort, ts1.get(1).sort
|
26
|
+
assert_equal %w(qux qip).sort, ts2.get(1).sort
|
27
|
+
end
|
28
|
+
|
29
|
+
def test_empty?
|
30
|
+
assert @ts.respond_to? :empty?
|
31
|
+
assert @ts.empty?
|
32
|
+
@ts.add(1, 'qux', 'qip')
|
33
|
+
refute @ts.empty?
|
34
|
+
@ts.delete(1)
|
35
|
+
assert @ts.empty?
|
36
|
+
end
|
37
|
+
|
38
|
+
def test_sets_tags_for_resource
|
39
|
+
assert_equal [], @ts.get(1)
|
40
|
+
@ts.set(1, 'foo', 'bar', 'baz')
|
41
|
+
assert_equal %w(bar baz foo).sort, @ts.get(1).sort
|
42
|
+
@ts.set(1, 'qux', 'qip')
|
43
|
+
assert_equal %w(qip qux).sort, @ts.get(1).sort
|
44
|
+
end
|
45
|
+
|
46
|
+
def test_sets_tags_when_empty
|
47
|
+
@ts.set(1, 'foo', 'bar', 'baz')
|
48
|
+
@ts.set(2, 'qux', 'qip')
|
49
|
+
assert_equal %w(bar baz foo).sort, @ts.get(1).sort
|
50
|
+
assert_equal %w(qip qux).sort, @ts.get(2).sort
|
51
|
+
@ts.set(1, *[])
|
52
|
+
assert_equal [], @ts.get(1)
|
53
|
+
assert_equal %w(qip qux).sort, @ts.get(2).sort
|
54
|
+
end
|
55
|
+
|
56
|
+
def test_deletes_all_tags_for_resource
|
57
|
+
@ts.set(1, 'foo', 'bar', 'baz')
|
58
|
+
@ts.set(2, 'qux', 'qip')
|
59
|
+
assert_equal %w(bar baz foo).sort, @ts.get(1).sort
|
60
|
+
assert_equal %w(qip qux).sort, @ts.get(2).sort
|
61
|
+
@ts.delete(1)
|
62
|
+
assert_equal [], @ts.get(1)
|
63
|
+
assert_equal %w(qip qux).sort, @ts.get(2).sort
|
64
|
+
end
|
65
|
+
|
66
|
+
def test_deletes_given_tags_for_resource
|
67
|
+
assert_equal [], @ts.get(1)
|
68
|
+
@ts.set(1, 'foo', 'bar', 'baz', 'qux', 'qip')
|
69
|
+
@ts.delete(1, 'qux', 'qip')
|
70
|
+
assert_equal %w(bar baz foo).sort, @ts.get(1).sort
|
71
|
+
end
|
72
|
+
|
73
|
+
def test_matches_tags
|
74
|
+
@ts.set(1, 'foo', 'bar', 'baz')
|
75
|
+
@ts.set(2, 'qux', 'qip')
|
76
|
+
|
77
|
+
assert @ts.matches(1, ['foo'])
|
78
|
+
assert @ts.matches(1, %w(bar baz))
|
79
|
+
assert @ts.matches(1, %w(bar baz foo))
|
80
|
+
refute @ts.matches(1, ['qux'])
|
81
|
+
refute @ts.matches(1, ['qip'])
|
82
|
+
|
83
|
+
refute @ts.matches(2, ['foo'])
|
84
|
+
refute @ts.matches(2, %w(bar baz))
|
85
|
+
refute @ts.matches(2, %w(bar baz foo))
|
86
|
+
assert @ts.matches(2, %w(qux qip))
|
87
|
+
assert @ts.matches(2, ['qux'])
|
88
|
+
assert @ts.matches(2, ['qip'])
|
89
|
+
end
|
90
|
+
|
91
|
+
def test_matches_exclude_tags
|
92
|
+
@ts.set(1, 'foo', 'bar', 'baz')
|
93
|
+
@ts.set(2, 'qux', 'qip')
|
94
|
+
|
95
|
+
refute @ts.matches(1, nil, ['foo'])
|
96
|
+
refute @ts.matches(1, [], ['foo'])
|
97
|
+
refute @ts.matches(1, [], %w(bar baz))
|
98
|
+
refute @ts.matches(1, [], %w(bar baz foo))
|
99
|
+
assert @ts.matches(1, [], ['qux'])
|
100
|
+
assert @ts.matches(1, [], ['qip'])
|
101
|
+
|
102
|
+
assert @ts.matches(2, nil, ['foo'])
|
103
|
+
assert @ts.matches(2, [], ['foo'])
|
104
|
+
assert @ts.matches(2, [], %w(bar baz))
|
105
|
+
assert @ts.matches(2, [], %w(bar baz foo))
|
106
|
+
refute @ts.matches(2, [], %w(qux qip))
|
107
|
+
refute @ts.matches(2, [], ['qux'])
|
108
|
+
refute @ts.matches(2, [], ['qip'])
|
109
|
+
end
|
110
|
+
|
111
|
+
def test_matches_include_and_exclude_tags
|
112
|
+
@ts.set(1, 'foo', 'bar', 'baz')
|
113
|
+
@ts.set(2, 'qux', 'qip')
|
114
|
+
|
115
|
+
refute @ts.matches(1, ['foo'], ['bar'])
|
116
|
+
refute @ts.matches(1, ['bar'], ['foo'])
|
117
|
+
|
118
|
+
assert @ts.matches(1, ['foo'], [])
|
119
|
+
assert @ts.matches(1, ['foo'], nil)
|
120
|
+
assert @ts.matches(1, ['foo'], ['qux'])
|
121
|
+
|
122
|
+
assert @ts.matches(2, ['qip'], ['foo'])
|
123
|
+
assert @ts.matches(2, ['qux'], %w(bar baz))
|
124
|
+
assert @ts.matches(2, ['qip'], %w(bar baz foo))
|
125
|
+
refute @ts.matches(2, ['qip'], %w(qux qip))
|
126
|
+
refute @ts.matches(2, ['qip'], ['qux'])
|
127
|
+
refute @ts.matches(2, ['qux'], ['qip'])
|
128
|
+
end
|
129
|
+
|
130
|
+
end
|
@@ -0,0 +1,106 @@
|
|
1
|
+
module TestsForWeightedGroups
|
2
|
+
|
3
|
+
def test_calls_each_content_set
|
4
|
+
expected = [
|
5
|
+
{resource: '6', similarity: 74.037},
|
6
|
+
{resource: '12', similarity: 55.5},
|
7
|
+
{resource: '9', similarity: 6.67},
|
8
|
+
{resource: '3', similarity: 4.0},
|
9
|
+
{resource: '21', similarity: 2.86},
|
10
|
+
{resource: '15', similarity: 2.86}
|
11
|
+
]
|
12
|
+
assert_equal expected, @weighted_group.similar_to(18)
|
13
|
+
end
|
14
|
+
|
15
|
+
def test_calls_each_content_set_with_limits
|
16
|
+
expected = [
|
17
|
+
{resource: '6', similarity: 74.037},
|
18
|
+
{resource: '12', similarity: 55.5},
|
19
|
+
{resource: '9', similarity: 6.67},
|
20
|
+
{resource: '3', similarity: 4.0},
|
21
|
+
{resource: '21', similarity: 2.86},
|
22
|
+
{resource: '15', similarity: 2.86}
|
23
|
+
]
|
24
|
+
assert_equal expected[0..0], @weighted_group.similar_to(18, 1)
|
25
|
+
assert_equal expected[0..2], @weighted_group.similar_to(18, 3)
|
26
|
+
assert_equal expected, @weighted_group.similar_to(18, 6)
|
27
|
+
assert_equal expected, @weighted_group.similar_to(18, 99)
|
28
|
+
end
|
29
|
+
|
30
|
+
def test_filters_include_recommendations
|
31
|
+
expected = [{resource: '15', similarity: 2.86}]
|
32
|
+
@weighted_group.tag_set = @tag_set
|
33
|
+
assert_equal expected, @weighted_group.filtered_similar_to(18, include: ['mod5'])
|
34
|
+
end
|
35
|
+
|
36
|
+
def test_filters_exclude_recommendations
|
37
|
+
expected = [
|
38
|
+
{resource: '6', similarity: 74.037},
|
39
|
+
{resource: '12', similarity: 55.5},
|
40
|
+
{resource: '9', similarity: 6.67},
|
41
|
+
{resource: '3', similarity: 4.0}
|
42
|
+
]
|
43
|
+
@weighted_group.tag_set = @tag_set
|
44
|
+
assert_equal expected, @weighted_group.filtered_similar_to(18, exclude: ['mod5', 'mod7'])
|
45
|
+
end
|
46
|
+
|
47
|
+
def test_filters_include_and_exclude_recommendations
|
48
|
+
expected = [
|
49
|
+
{resource: '16', similarity: 0.8},
|
50
|
+
{resource: '4', similarity: 0.667},
|
51
|
+
{resource: '12', similarity: 0.333}
|
52
|
+
]
|
53
|
+
@weighted_group.tag_set = @tag_set
|
54
|
+
assert_equal expected, @weighted_group.filtered_similar_to(8, include: ['mod4'], exclude: ['mod5'])
|
55
|
+
end
|
56
|
+
|
57
|
+
def test_filters_include_and_exclude_recommendations_and_limits
|
58
|
+
expected = [
|
59
|
+
{resource: '16', similarity: 0.8},
|
60
|
+
{resource: '4', similarity: 0.667},
|
61
|
+
{resource: '12', similarity: 0.333}
|
62
|
+
]
|
63
|
+
@weighted_group.tag_set = @tag_set
|
64
|
+
assert_equal expected[0..0], @weighted_group.filtered_similar_to(8, include: ['mod4'], exclude: ['mod5'], limit: 1)
|
65
|
+
assert_equal expected[0..1], @weighted_group.filtered_similar_to(8, include: ['mod4'], exclude: ['mod5'], limit: 2)
|
66
|
+
assert_equal expected, @weighted_group.filtered_similar_to(8, include: ['mod4'], exclude: ['mod5'], limit: 3)
|
67
|
+
assert_equal expected, @weighted_group.filtered_similar_to(8, include: ['mod4'], exclude: ['mod5'], limit: 99)
|
68
|
+
end
|
69
|
+
|
70
|
+
def test_similar_to_mutliple_items
|
71
|
+
expected = [
|
72
|
+
{resource: '12', similarity: 78.437},
|
73
|
+
{resource: '18', similarity: 78.037},
|
74
|
+
{resource: '9', similarity: 11.67},
|
75
|
+
{resource: '21', similarity: 9.0},
|
76
|
+
{resource: '15', similarity: 9.0},
|
77
|
+
{resource: '6', similarity: 6.67},
|
78
|
+
{resource: '3', similarity: 6.67},
|
79
|
+
{resource: '8', similarity: 0.667},
|
80
|
+
{resource: '16', similarity: 0.5},
|
81
|
+
{resource: '20', similarity: 0.4}
|
82
|
+
]
|
83
|
+
@weighted_group.tag_set = @tag_set
|
84
|
+
assert_equal expected, @weighted_group.similar_to([3, 4, 5, 6, 7])
|
85
|
+
end
|
86
|
+
|
87
|
+
def test_filtered_similar_to_mutliple_items
|
88
|
+
expected = [
|
89
|
+
{resource: '12', similarity: 78.437},
|
90
|
+
{resource: '8', similarity: 0.667},
|
91
|
+
{resource: '16', similarity: 0.5},
|
92
|
+
]
|
93
|
+
|
94
|
+
@weighted_group.tag_set = @tag_set
|
95
|
+
assert_equal expected, @weighted_group.filtered_similar_to([3, 4, 5, 6, 7], include: ['mod4'], exclude: ['mod5'])
|
96
|
+
end
|
97
|
+
|
98
|
+
def test_weightings_affect_scores
|
99
|
+
skip
|
100
|
+
end
|
101
|
+
|
102
|
+
def test_precalculates
|
103
|
+
skip
|
104
|
+
end
|
105
|
+
|
106
|
+
end
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: commendo
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version:
|
4
|
+
version: 2.0.0
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Rob Styles
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2015-12-
|
11
|
+
date: 2015-12-18 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: redis
|
@@ -24,6 +24,20 @@ dependencies:
|
|
24
24
|
- - ">="
|
25
25
|
- !ruby/object:Gem::Version
|
26
26
|
version: '0'
|
27
|
+
- !ruby/object:Gem::Dependency
|
28
|
+
name: mysql2
|
29
|
+
requirement: !ruby/object:Gem::Requirement
|
30
|
+
requirements:
|
31
|
+
- - ">="
|
32
|
+
- !ruby/object:Gem::Version
|
33
|
+
version: '0'
|
34
|
+
type: :runtime
|
35
|
+
prerelease: false
|
36
|
+
version_requirements: !ruby/object:Gem::Requirement
|
37
|
+
requirements:
|
38
|
+
- - ">="
|
39
|
+
- !ruby/object:Gem::Version
|
40
|
+
version: '0'
|
27
41
|
- !ruby/object:Gem::Dependency
|
28
42
|
name: progressbar
|
29
43
|
requirement: !ruby/object:Gem::Requirement
|
@@ -38,6 +52,20 @@ dependencies:
|
|
38
52
|
- - ">="
|
39
53
|
- !ruby/object:Gem::Version
|
40
54
|
version: '0'
|
55
|
+
- !ruby/object:Gem::Dependency
|
56
|
+
name: slop
|
57
|
+
requirement: !ruby/object:Gem::Requirement
|
58
|
+
requirements:
|
59
|
+
- - ">="
|
60
|
+
- !ruby/object:Gem::Version
|
61
|
+
version: '0'
|
62
|
+
type: :runtime
|
63
|
+
prerelease: false
|
64
|
+
version_requirements: !ruby/object:Gem::Requirement
|
65
|
+
requirements:
|
66
|
+
- - ">="
|
67
|
+
- !ruby/object:Gem::Version
|
68
|
+
version: '0'
|
41
69
|
- !ruby/object:Gem::Dependency
|
42
70
|
name: bundler
|
43
71
|
requirement: !ruby/object:Gem::Requirement
|
@@ -108,16 +136,20 @@ dependencies:
|
|
108
136
|
- - "~>"
|
109
137
|
- !ruby/object:Gem::Version
|
110
138
|
version: 5.0.8
|
111
|
-
description: A Jaccard-similarity recommender using Redis sets
|
139
|
+
description: A Jaccard-similarity recommender using Redis sets or MySQL
|
112
140
|
email:
|
113
141
|
- rob.styles@dynamicorange.com
|
114
142
|
executables:
|
143
|
+
- commendo-create-mysql-db
|
144
|
+
- commendo-create.sql
|
115
145
|
- commendo-find-identical-pairs
|
116
146
|
- commendo-load
|
117
147
|
- commendo-load-tags-tsv
|
118
148
|
- commendo-load-tsv
|
149
|
+
- commendo-load-tsv-mysql.rb
|
119
150
|
- commendo-recommendations-distribution
|
120
151
|
- commendo-similarity-distribution
|
152
|
+
- commendo-time-mysql.rb
|
121
153
|
extensions: []
|
122
154
|
extra_rdoc_files: []
|
123
155
|
files:
|
@@ -127,23 +159,44 @@ files:
|
|
127
159
|
- LICENSE.txt
|
128
160
|
- README.md
|
129
161
|
- Rakefile
|
162
|
+
- bin/commendo-create-mysql-db
|
163
|
+
- bin/commendo-create.sql
|
130
164
|
- bin/commendo-find-identical-pairs
|
131
165
|
- bin/commendo-load
|
132
166
|
- bin/commendo-load-tags-tsv
|
133
167
|
- bin/commendo-load-tsv
|
168
|
+
- bin/commendo-load-tsv-mysql.rb
|
134
169
|
- bin/commendo-recommendations-distribution
|
135
170
|
- bin/commendo-similarity-distribution
|
171
|
+
- bin/commendo-time-mysql.rb
|
136
172
|
- commendo.gemspec
|
137
173
|
- lib/commendo.rb
|
174
|
+
- lib/commendo/configuration.rb
|
138
175
|
- lib/commendo/content_set.rb
|
139
|
-
- lib/commendo/
|
140
|
-
- lib/commendo/
|
176
|
+
- lib/commendo/mysql-backed/content_set.rb
|
177
|
+
- lib/commendo/mysql-backed/tag_set.rb
|
178
|
+
- lib/commendo/mysql-backed/weighted_group.rb
|
179
|
+
- lib/commendo/redis-backed/content_set.rb
|
180
|
+
- lib/commendo/redis-backed/pair_comparison.lua
|
181
|
+
- lib/commendo/redis-backed/similarity.lua
|
182
|
+
- lib/commendo/redis-backed/tag_set.rb
|
183
|
+
- lib/commendo/redis-backed/weighted_group.rb
|
141
184
|
- lib/commendo/tag_set.rb
|
142
185
|
- lib/commendo/version.rb
|
143
186
|
- lib/commendo/weighted_group.rb
|
144
|
-
-
|
145
|
-
-
|
146
|
-
-
|
187
|
+
- lib/mysql2/client.rb
|
188
|
+
- model 2.mwb
|
189
|
+
- sql_model.mwb
|
190
|
+
- test/configuration_test.rb
|
191
|
+
- test/mysql_content_set_test.rb
|
192
|
+
- test/mysql_tag_set_test.rb
|
193
|
+
- test/mysql_weighted_group_test.rb
|
194
|
+
- test/redis_content_set_test.rb
|
195
|
+
- test/redis_tag_set_test.rb
|
196
|
+
- test/redis_weighted_group_test.rb
|
197
|
+
- test/tests_for_content_sets.rb
|
198
|
+
- test/tests_for_tag_sets.rb
|
199
|
+
- test/tests_for_weighted_groups.rb
|
147
200
|
homepage: ''
|
148
201
|
licenses:
|
149
202
|
- MIT
|
@@ -167,9 +220,16 @@ rubyforge_project:
|
|
167
220
|
rubygems_version: 2.5.0
|
168
221
|
signing_key:
|
169
222
|
specification_version: 4
|
170
|
-
summary: A Jaccard-similarity recommender using Redis sets
|
223
|
+
summary: A Jaccard-similarity recommender using Redis sets or MySQL
|
171
224
|
test_files:
|
172
|
-
- test/
|
173
|
-
- test/
|
174
|
-
- test/
|
225
|
+
- test/configuration_test.rb
|
226
|
+
- test/mysql_content_set_test.rb
|
227
|
+
- test/mysql_tag_set_test.rb
|
228
|
+
- test/mysql_weighted_group_test.rb
|
229
|
+
- test/redis_content_set_test.rb
|
230
|
+
- test/redis_tag_set_test.rb
|
231
|
+
- test/redis_weighted_group_test.rb
|
232
|
+
- test/tests_for_content_sets.rb
|
233
|
+
- test/tests_for_tag_sets.rb
|
234
|
+
- test/tests_for_weighted_groups.rb
|
175
235
|
has_rdoc:
|
data/test/content_set_test.rb
DELETED
@@ -1,408 +0,0 @@
|
|
1
|
-
gem 'minitest'
|
2
|
-
require 'minitest/autorun'
|
3
|
-
require 'minitest/pride'
|
4
|
-
require 'minitest/mock'
|
5
|
-
require 'mocha/setup'
|
6
|
-
require 'commendo'
|
7
|
-
|
8
|
-
module Commendo
|
9
|
-
|
10
|
-
class ContentSetTest < Minitest::Test
|
11
|
-
|
12
|
-
def setup
|
13
|
-
@redis = Redis.new(db: 15)
|
14
|
-
@redis.flushdb
|
15
|
-
@key_base = 'CommendoTests'
|
16
|
-
@cs = ContentSet.new(@redis, @key_base)
|
17
|
-
end
|
18
|
-
|
19
|
-
def test_gives_similarity_key_for_resource
|
20
|
-
key_base = 'CommendoTestsFooBarBaz'
|
21
|
-
cs = ContentSet.new(@redis, key_base)
|
22
|
-
assert_equal 'CommendoTestsFooBarBaz:similar:resource-1', cs.similarity_key('resource-1')
|
23
|
-
end
|
24
|
-
|
25
|
-
def test_recommends_when_added
|
26
|
-
@cs.add('resource-1', 'group-1', 'group-2')
|
27
|
-
@cs.add('resource-2', 'group-1')
|
28
|
-
@cs.add('resource-3', 'group-1', 'group-2')
|
29
|
-
@cs.add('resource-4', 'group-2')
|
30
|
-
@cs.calculate_similarity
|
31
|
-
expected = [
|
32
|
-
{resource: 'resource-3', similarity: 1.0},
|
33
|
-
{resource: 'resource-4', similarity: 0.667},
|
34
|
-
{resource: 'resource-2', similarity: 0.667}
|
35
|
-
]
|
36
|
-
assert_equal expected, @cs.similar_to('resource-1')
|
37
|
-
end
|
38
|
-
|
39
|
-
def test_recommends_limited_by_number
|
40
|
-
@cs.add('resource-1', 'group-1', 'group-2')
|
41
|
-
@cs.add('resource-2', 'group-1')
|
42
|
-
@cs.add('resource-3', 'group-1', 'group-2')
|
43
|
-
@cs.add('resource-4', 'group-2')
|
44
|
-
@cs.calculate_similarity
|
45
|
-
expected = [
|
46
|
-
{resource: 'resource-3', similarity: 1.0},
|
47
|
-
{resource: 'resource-4', similarity: 0.667},
|
48
|
-
{resource: 'resource-2', similarity: 0.667}
|
49
|
-
]
|
50
|
-
assert_equal expected[0..0], @cs.similar_to('resource-1', 1)
|
51
|
-
assert_equal expected[0..1], @cs.similar_to('resource-1', 2)
|
52
|
-
assert_equal expected, @cs.similar_to('resource-1', 3)
|
53
|
-
assert_equal expected, @cs.similar_to('resource-1', 99)
|
54
|
-
end
|
55
|
-
|
56
|
-
def test_recommends_when_added_with_scores
|
57
|
-
@cs.add('resource-1', ['group-1', 2], ['group-2', 2])
|
58
|
-
@cs.add('resource-2', ['group-1', 7])
|
59
|
-
@cs.add('resource-3', ['group-1', 2], ['group-2', 2])
|
60
|
-
@cs.add('resource-4', ['group-2', 3])
|
61
|
-
@cs.calculate_similarity
|
62
|
-
expected = [
|
63
|
-
{resource: 'resource-3', similarity: 1.0},
|
64
|
-
{resource: 'resource-2', similarity: 0.818},
|
65
|
-
{resource: 'resource-4', similarity: 0.714}
|
66
|
-
]
|
67
|
-
assert_equal expected, @cs.similar_to('resource-1')
|
68
|
-
end
|
69
|
-
|
70
|
-
def test_recommends_with_large_number_of_groups
|
71
|
-
(0..3000).each do |i|
|
72
|
-
@cs.add('resource-1', ["group-#{i}", i/100.0], ["group-#{i+1}", i/20.0])
|
73
|
-
@cs.add('resource-9', ["group-#{i}", i/100.0], ["group-#{i+1}", i/20.0])
|
74
|
-
end
|
75
|
-
@cs.calculate_similarity
|
76
|
-
expected = [
|
77
|
-
{resource: 'resource-9', similarity: 1.0}
|
78
|
-
]
|
79
|
-
assert_equal expected, @cs.similar_to('resource-1')
|
80
|
-
end
|
81
|
-
|
82
|
-
def test_recommends_when_extra_scores_added
|
83
|
-
test_recommends_when_added_with_scores #sets up the content set
|
84
|
-
@cs.add('resource-3', ['group-1', 1], ['group-3', 2])
|
85
|
-
@cs.add('resource-4', ['group-2', 1])
|
86
|
-
@cs.add_by_group('group-1', ['newource-9', 100], 'resource-2', 'resource-3')
|
87
|
-
@cs.add_by_group('group-2', 'resource-1', 'resource-3', 'resource-4')
|
88
|
-
@cs.calculate_similarity
|
89
|
-
expected = [
|
90
|
-
{resource: 'newource-9', similarity: 1.0},
|
91
|
-
{resource: 'resource-1', similarity: 0.769},
|
92
|
-
{resource: 'resource-3', similarity: 0.706}
|
93
|
-
]
|
94
|
-
assert_equal expected, @cs.similar_to('resource-2')
|
95
|
-
end
|
96
|
-
|
97
|
-
def test_recommends_when_added_by_group
|
98
|
-
@cs.add_by_group('group-1', 'resource-1', 'resource-2', 'resource-3')
|
99
|
-
@cs.add_by_group('group-2', 'resource-1', 'resource-3', 'resource-4')
|
100
|
-
@cs.calculate_similarity
|
101
|
-
expected = [
|
102
|
-
{resource: 'resource-3', similarity: 1.0},
|
103
|
-
{resource: 'resource-4', similarity: 0.667},
|
104
|
-
{resource: 'resource-2', similarity: 0.667}
|
105
|
-
]
|
106
|
-
assert_equal expected, @cs.similar_to('resource-1')
|
107
|
-
end
|
108
|
-
|
109
|
-
def test_recommends_when_added_by_group_with_scores
|
110
|
-
@cs.add_by_group('group-1', ['resource-1', 2], ['resource-2', 3], ['resource-3', 7])
|
111
|
-
@cs.add_by_group('group-2', ['resource-1', 2], ['resource-3', 3], ['resource-4', 5])
|
112
|
-
@cs.calculate_similarity
|
113
|
-
expected = [
|
114
|
-
{resource: 'resource-3', similarity: 1.0},
|
115
|
-
{resource: 'resource-4', similarity: 0.778},
|
116
|
-
{resource: 'resource-2', similarity: 0.714}
|
117
|
-
]
|
118
|
-
assert_equal expected, @cs.similar_to('resource-1')
|
119
|
-
end
|
120
|
-
|
121
|
-
def test_recommendations_are_isolated_by_key_base
|
122
|
-
skip
|
123
|
-
end
|
124
|
-
|
125
|
-
def test_recommendations_are_isolated_by_redis_db
|
126
|
-
skip
|
127
|
-
end
|
128
|
-
|
129
|
-
def test_calculates_similarity_scores
|
130
|
-
(3..23).each do |group|
|
131
|
-
(3..23).each do |res|
|
132
|
-
@cs.add_by_group(group, res) if res % group == 0
|
133
|
-
end
|
134
|
-
end
|
135
|
-
@cs.calculate_similarity
|
136
|
-
expected = [
|
137
|
-
{resource: '9', similarity: 0.667},
|
138
|
-
{resource: '6', similarity: 0.667},
|
139
|
-
{resource: '12', similarity: 0.5},
|
140
|
-
{resource: '3', similarity: 0.4},
|
141
|
-
{resource: '21', similarity: 0.286},
|
142
|
-
{resource: '15', similarity: 0.286}
|
143
|
-
]
|
144
|
-
assert_equal expected, @cs.similar_to(18)
|
145
|
-
end
|
146
|
-
|
147
|
-
def test_calculates_with_threshold
|
148
|
-
(3..23).each do |group|
|
149
|
-
(3..23).each do |res|
|
150
|
-
@cs.add_by_group(group, res) if res % group == 0
|
151
|
-
end
|
152
|
-
end
|
153
|
-
@cs.calculate_similarity(0.4)
|
154
|
-
expected = [
|
155
|
-
{resource: '9', similarity: 0.667},
|
156
|
-
{resource: '6', similarity: 0.667},
|
157
|
-
{resource: '12', similarity: 0.5}
|
158
|
-
]
|
159
|
-
assert_equal expected, @cs.similar_to(18)
|
160
|
-
end
|
161
|
-
|
162
|
-
def test_calculate_copes_with_missing_resource
|
163
|
-
@cs.calculate_similarity_for_resource('999999999999', 0.1)
|
164
|
-
end
|
165
|
-
|
166
|
-
def test_calculate_yields_after_each
|
167
|
-
(3..23).each do |group|
|
168
|
-
(3..23).each do |res|
|
169
|
-
@cs.add_by_group(group, res) if res % group == 0
|
170
|
-
end
|
171
|
-
end
|
172
|
-
expected_keys = ['CommendoTests:resources:3', 'CommendoTests:resources:4', 'CommendoTests:resources:5', 'CommendoTests:resources:6', 'CommendoTests:resources:7', 'CommendoTests:resources:8', 'CommendoTests:resources:9', 'CommendoTests:resources:10', 'CommendoTests:resources:11', 'CommendoTests:resources:12', 'CommendoTests:resources:13', 'CommendoTests:resources:14', 'CommendoTests:resources:15', 'CommendoTests:resources:16', 'CommendoTests:resources:17', 'CommendoTests:resources:18', 'CommendoTests:resources:19', 'CommendoTests:resources:20', 'CommendoTests:resources:21', 'CommendoTests:resources:22', 'CommendoTests:resources:23']
|
173
|
-
actual_keys = []
|
174
|
-
@cs.calculate_similarity { |key, index, total|
|
175
|
-
actual_keys << key
|
176
|
-
}
|
177
|
-
assert_equal expected_keys.sort, actual_keys.sort
|
178
|
-
end
|
179
|
-
|
180
|
-
def test_deletes_resource_from_everywhere
|
181
|
-
(3..23).each do |group|
|
182
|
-
(3..23).each do |res|
|
183
|
-
@cs.add_by_group(group, res) if res % group == 0
|
184
|
-
end
|
185
|
-
end
|
186
|
-
@cs.calculate_similarity
|
187
|
-
assert similar_to(@cs, 18, 12)
|
188
|
-
|
189
|
-
@cs.delete(12)
|
190
|
-
assert_equal [], @cs.similar_to(12)
|
191
|
-
refute similar_to(@cs, 18, 12)
|
192
|
-
|
193
|
-
@cs.calculate_similarity
|
194
|
-
assert_equal [], @cs.similar_to(12)
|
195
|
-
refute similar_to(@cs, 18, 12)
|
196
|
-
end
|
197
|
-
|
198
|
-
def test_remove_from_groups
|
199
|
-
(3..23).each do |group|
|
200
|
-
(3..23).each do |res|
|
201
|
-
@cs.add(res, group) if res % group == 0
|
202
|
-
end
|
203
|
-
end
|
204
|
-
resource = 20
|
205
|
-
assert_equal ['4','5','10','20'].sort!, @cs.groups(resource).sort!
|
206
|
-
@cs.remove_from_groups(resource, 10)
|
207
|
-
assert_equal ['4','5','20'].sort!, @cs.groups(resource).sort!
|
208
|
-
@cs.remove_from_groups(resource, 4)
|
209
|
-
assert_equal ['5','20'].sort!, @cs.groups(resource).sort!
|
210
|
-
end
|
211
|
-
|
212
|
-
def test_remove_causes_similarity_to_change_when_recalculated
|
213
|
-
(3..23).each do |group|
|
214
|
-
(3..23).each do |res|
|
215
|
-
@cs.add(res, group) if res % group == 0
|
216
|
-
end
|
217
|
-
end
|
218
|
-
@cs.calculate_similarity
|
219
|
-
assert similar_to(@cs, 18, 12)
|
220
|
-
@cs.remove_from_groups(18, 6, 3)
|
221
|
-
@cs.calculate_similarity
|
222
|
-
refute similar_to(@cs, 18, 12)
|
223
|
-
end
|
224
|
-
|
225
|
-
def test_accepts_incremental_updates
|
226
|
-
(3..23).each do |group|
|
227
|
-
(3..23).each do |res|
|
228
|
-
@cs.add(res, group) if res % group == 0
|
229
|
-
end
|
230
|
-
end
|
231
|
-
@cs.calculate_similarity
|
232
|
-
assert similar_to(@cs, 18, 12)
|
233
|
-
refute similar_to(@cs, 10, 12)
|
234
|
-
|
235
|
-
@cs.add_and_calculate(12, 'foo', true)
|
236
|
-
@cs.add_and_calculate(10, 'foo', true)
|
237
|
-
assert similar_to(@cs, 10, 12)
|
238
|
-
end
|
239
|
-
|
240
|
-
def test_remove_and_calculate
|
241
|
-
(3..23).each do |group|
|
242
|
-
(3..23).each do |res|
|
243
|
-
@cs.add(res, group) if res % group == 0
|
244
|
-
end
|
245
|
-
end
|
246
|
-
@cs.calculate_similarity
|
247
|
-
assert similar_to(@cs, 18, 12)
|
248
|
-
@cs.remove_from_groups_and_calculate(18, 6, 3)
|
249
|
-
refute similar_to(@cs, 18, 12)
|
250
|
-
end
|
251
|
-
|
252
|
-
def test_filters_include_by_tag_collection
|
253
|
-
ts = TagSet.new(@redis, "#{@key_base}:tags")
|
254
|
-
@cs = ContentSet.new(@redis, @key_base, ts)
|
255
|
-
(3..23).each do |group|
|
256
|
-
(3..23).each do |res|
|
257
|
-
@cs.add(res, group) if res % group == 0
|
258
|
-
ts.add(res, 'mod3') if res.modulo(3).zero?
|
259
|
-
ts.add(res, 'mod4') if res.modulo(4).zero?
|
260
|
-
ts.add(res, 'mod5') if res.modulo(5).zero?
|
261
|
-
end
|
262
|
-
end
|
263
|
-
@cs.calculate_similarity
|
264
|
-
|
265
|
-
actual = @cs.filtered_similar_to(10, include: ['mod5'])
|
266
|
-
assert_equal 3, actual.length
|
267
|
-
assert contains_resource('5', actual)
|
268
|
-
assert contains_resource('15', actual)
|
269
|
-
assert contains_resource('20', actual)
|
270
|
-
|
271
|
-
end
|
272
|
-
|
273
|
-
def test_filters_include_by_tag_collection_and_limit
|
274
|
-
ts = TagSet.new(@redis, "#{@key_base}:tags")
|
275
|
-
@cs = ContentSet.new(@redis, @key_base, ts)
|
276
|
-
(3..23).each do |group|
|
277
|
-
(3..23).each do |res|
|
278
|
-
@cs.add(res, group) if res % group == 0
|
279
|
-
ts.add(res, 'mod3') if res.modulo(3).zero?
|
280
|
-
ts.add(res, 'mod4') if res.modulo(4).zero?
|
281
|
-
ts.add(res, 'mod5') if res.modulo(5).zero?
|
282
|
-
end
|
283
|
-
end
|
284
|
-
@cs.calculate_similarity
|
285
|
-
|
286
|
-
actual = @cs.filtered_similar_to(10, include: ['mod5'], limit: 2)
|
287
|
-
assert_equal 2, actual.length
|
288
|
-
assert contains_resource('5', actual)
|
289
|
-
#assert contains_resource('15', actual)
|
290
|
-
assert contains_resource('20', actual)
|
291
|
-
|
292
|
-
end
|
293
|
-
|
294
|
-
def test_filters_exclude_by_tag_collection
|
295
|
-
ts = TagSet.new(@redis, "#{@key_base}:tags")
|
296
|
-
@cs = ContentSet.new(@redis, @key_base, ts)
|
297
|
-
(3..23).each do |group|
|
298
|
-
(3..23).each do |res|
|
299
|
-
@cs.add(res, group) if res % group == 0
|
300
|
-
ts.add(res, 'mod3') if res.modulo(3).zero?
|
301
|
-
ts.add(res, 'mod4') if res.modulo(4).zero?
|
302
|
-
ts.add(res, 'mod5') if res.modulo(5).zero?
|
303
|
-
end
|
304
|
-
end
|
305
|
-
@cs.calculate_similarity
|
306
|
-
|
307
|
-
actual = @cs.filtered_similar_to(10, exclude: ['mod3'])
|
308
|
-
assert_equal 2, actual.length
|
309
|
-
assert contains_resource('5', actual)
|
310
|
-
assert contains_resource('20', actual)
|
311
|
-
refute contains_resource('15', actual)
|
312
|
-
end
|
313
|
-
|
314
|
-
def test_filters_includes_and_exclude_by_tag_collection
|
315
|
-
ts = TagSet.new(@redis, "#{@key_base}:tags")
|
316
|
-
@cs = ContentSet.new(@redis, @key_base, ts)
|
317
|
-
#Build some test data
|
318
|
-
(3..23).each do |group|
|
319
|
-
(3..23).each do |res|
|
320
|
-
@cs.add(res, group) if res % group == 0
|
321
|
-
ts.add(res, 'mod3') if res.modulo(3).zero?
|
322
|
-
ts.add(res, 'mod4') if res.modulo(4).zero?
|
323
|
-
ts.add(res, 'mod5') if res.modulo(5).zero?
|
324
|
-
end
|
325
|
-
end
|
326
|
-
@cs.calculate_similarity
|
327
|
-
|
328
|
-
actual = @cs.filtered_similar_to(12, include: ['mod4'], exclude: ['mod3', 'mod5'])
|
329
|
-
assert_equal 3, actual.length
|
330
|
-
|
331
|
-
refute contains_resource('6', actual)
|
332
|
-
refute contains_resource('18', actual)
|
333
|
-
assert contains_resource('4', actual)
|
334
|
-
refute contains_resource('3', actual)
|
335
|
-
refute contains_resource('9', actual)
|
336
|
-
assert contains_resource('8', actual)
|
337
|
-
refute contains_resource('21', actual)
|
338
|
-
assert contains_resource('16', actual)
|
339
|
-
refute contains_resource('15', actual)
|
340
|
-
refute contains_resource('20', actual)
|
341
|
-
end
|
342
|
-
|
343
|
-
def test_recommends_for_many
|
344
|
-
ts = TagSet.new(@redis, "#{@key_base}:tags")
|
345
|
-
@cs = ContentSet.new(@redis, @key_base, ts)
|
346
|
-
(3..23).each do |group|
|
347
|
-
(3..23).each do |res|
|
348
|
-
@cs.add(res, group) if res % group == 0
|
349
|
-
ts.add(res, 'mod3') if res.modulo(3).zero?
|
350
|
-
ts.add(res, 'mod4') if res.modulo(4).zero?
|
351
|
-
ts.add(res, 'mod5') if res.modulo(5).zero?
|
352
|
-
end
|
353
|
-
end
|
354
|
-
@cs.calculate_similarity
|
355
|
-
expected = [
|
356
|
-
{resource: '18', similarity: 1.834},
|
357
|
-
{resource: '3', similarity: 1.734},
|
358
|
-
{resource: '6', similarity: 1.167},
|
359
|
-
{resource: '21', similarity: 1.086},
|
360
|
-
{resource: '15', similarity: 1.086},
|
361
|
-
{resource: '12', similarity: 1.0},
|
362
|
-
{resource: '9', similarity: 0.833},
|
363
|
-
{resource: '4', similarity: 0.4},
|
364
|
-
{resource: '8', similarity: 0.333},
|
365
|
-
{resource: '16', similarity: 0.286},
|
366
|
-
{resource: '20', similarity: 0.25}
|
367
|
-
]
|
368
|
-
actual = @cs.similar_to([12, 6, 9])
|
369
|
-
assert_equal expected, actual
|
370
|
-
#, include: ['mod4'], exclude: ['mod3', 'mod5']
|
371
|
-
end
|
372
|
-
|
373
|
-
def test_recommends_for_many_applies_filters
|
374
|
-
ts = TagSet.new(@redis, "#{@key_base}:tags")
|
375
|
-
@cs = ContentSet.new(@redis, @key_base, ts)
|
376
|
-
(3..23).each do |group|
|
377
|
-
(3..23).each do |res|
|
378
|
-
@cs.add(res, group) if res % group == 0
|
379
|
-
ts.add(res, 'mod3') if res.modulo(3).zero?
|
380
|
-
ts.add(res, 'mod4') if res.modulo(4).zero?
|
381
|
-
ts.add(res, 'mod5') if res.modulo(5).zero?
|
382
|
-
end
|
383
|
-
end
|
384
|
-
@cs.calculate_similarity
|
385
|
-
actual = @cs.filtered_similar_to([12, 6, 9], include: ['mod4'], exclude: ['mod3', 'mod5'])
|
386
|
-
refute contains_resource('6', actual)
|
387
|
-
refute contains_resource('18', actual)
|
388
|
-
assert contains_resource('4', actual)
|
389
|
-
refute contains_resource('3', actual)
|
390
|
-
refute contains_resource('9', actual)
|
391
|
-
assert contains_resource('8', actual)
|
392
|
-
refute contains_resource('21', actual)
|
393
|
-
assert contains_resource('16', actual)
|
394
|
-
refute contains_resource('15', actual)
|
395
|
-
refute contains_resource('20', actual)
|
396
|
-
end
|
397
|
-
|
398
|
-
def similar_to(cs, resource, similar)
|
399
|
-
contains_resource(similar, cs.similar_to(resource))
|
400
|
-
end
|
401
|
-
|
402
|
-
def contains_resource(resource, similarities)
|
403
|
-
!similarities.select { |sim| sim[:resource] == "#{resource}" }.empty?
|
404
|
-
end
|
405
|
-
|
406
|
-
end
|
407
|
-
|
408
|
-
end
|