otherinbox-classifier 1.3.1.1 → 1.3.1.2.20121218
Sign up to get free protection for your applications and to get access to all the features.
- data/lib/classifier/bayes.rb +37 -7
- data/lib/classifier/extensions/word_hash.rb +2 -0
- metadata +37 -42
data/lib/classifier/bayes.rb
CHANGED
@@ -28,6 +28,7 @@ class Bayes
|
|
28
28
|
@categories[category][word] += count
|
29
29
|
@total_words += count
|
30
30
|
end
|
31
|
+
reset_correct_counts!
|
31
32
|
end
|
32
33
|
|
33
34
|
#
|
@@ -55,6 +56,7 @@ class Bayes
|
|
55
56
|
@total_words -= count
|
56
57
|
end
|
57
58
|
end
|
59
|
+
reset_correct_counts!
|
58
60
|
end
|
59
61
|
|
60
62
|
#
|
@@ -80,24 +82,28 @@ class Bayes
|
|
80
82
|
myclassify_with_word_hash(text.word_hash)
|
81
83
|
end
|
82
84
|
|
83
|
-
|
84
|
-
def myclassify_with_word_hash(word_hash)
|
85
|
+
def myclassify_with_word_hash(word_hash, debugging_info = nil)
|
85
86
|
member_term_count = @categories[:Member].size
|
86
87
|
nonmember_term_count = @categories[:"Not member"].size
|
88
|
+
|
87
89
|
term_count = member_term_count + nonmember_term_count
|
88
90
|
score = 0
|
89
91
|
word_hash.each do |word, count|
|
90
92
|
# count of words in each category
|
91
|
-
member_count = @categories[:Member][word].to_i + 1
|
92
|
-
nonmember_count = @categories[:"Not member"][word].to_i + 1
|
93
|
-
next if member_count
|
93
|
+
member_count = @categories[:Member][word].to_i + 0.1
|
94
|
+
nonmember_count = @categories[:"Not member"][word].to_i + 0.1
|
95
|
+
next if member_count == 0.1 && nonmember_count == 0.1
|
94
96
|
|
95
97
|
# find relative prob word is in class -- p(w|c)
|
96
|
-
word_member_p = (member_count) / (
|
97
|
-
word_nonmember_p = (nonmember_count) / (
|
98
|
+
word_member_p = (member_count) / (total_member_count_correct + term_count).to_f
|
99
|
+
word_nonmember_p = (nonmember_count) / (total_nonmember_count_correct + term_count).to_f
|
98
100
|
|
99
101
|
word_pr = Math.log(word_member_p / word_nonmember_p)
|
100
102
|
score += word_pr * count
|
103
|
+
if debugging_info
|
104
|
+
debugging_info[word] = word_pr * count
|
105
|
+
end
|
106
|
+
#print "#{word_pr * count}: #{word}\n"
|
101
107
|
end
|
102
108
|
if score > 0
|
103
109
|
return "Member", score
|
@@ -155,10 +161,34 @@ class Bayes
|
|
155
161
|
def add_category(category)
|
156
162
|
@categories[category.prepare_category_name] = Hash.new
|
157
163
|
end
|
164
|
+
|
165
|
+
def remove_low_frequency_words(threshold = 5)
|
166
|
+
@categories.each do |_, word_counts|
|
167
|
+
word_counts.to_a.each do |word, count|
|
168
|
+
if count < threshold
|
169
|
+
word_counts.delete(word)
|
170
|
+
end
|
171
|
+
end
|
172
|
+
end
|
173
|
+
reset_correct_counts!
|
174
|
+
end
|
158
175
|
|
159
176
|
alias append_category add_category
|
160
177
|
|
161
178
|
private
|
179
|
+
def reset_correct_counts!
|
180
|
+
@total_member_count_correct = nil
|
181
|
+
@total_nonmember_count_correct = nil
|
182
|
+
end
|
183
|
+
|
184
|
+
def total_member_count_correct
|
185
|
+
@total_member_count_correct ||= @categories[:Member].values.inject(0) {|sum, element| sum+element}
|
186
|
+
end
|
187
|
+
|
188
|
+
def total_nonmember_count_correct
|
189
|
+
@total_nonmember_count_correct ||= @categories[:"Not member"].values.inject(0) {|sum, element| sum+element}
|
190
|
+
end
|
191
|
+
|
162
192
|
def total_member_count
|
163
193
|
@total_member_count ||= @categories[:Member].values.inject(0) {|sum, element| sum+element}
|
164
194
|
end
|
metadata
CHANGED
@@ -1,85 +1,80 @@
|
|
1
|
-
--- !ruby/object:Gem::Specification
|
1
|
+
--- !ruby/object:Gem::Specification
|
2
2
|
name: otherinbox-classifier
|
3
|
-
version: !ruby/object:Gem::Version
|
4
|
-
version: 1.3.1.
|
3
|
+
version: !ruby/object:Gem::Version
|
4
|
+
version: 1.3.1.2.20121218
|
5
|
+
prerelease:
|
5
6
|
platform: ruby
|
6
|
-
authors:
|
7
|
+
authors:
|
7
8
|
- Lucas Carlson
|
8
9
|
autorequire: classifier
|
9
10
|
bindir: bin
|
10
|
-
cert_chain:
|
11
|
-
date: 2008-01-19 00:00:00
|
12
|
-
|
13
|
-
|
14
|
-
- !ruby/object:Gem::Dependency
|
11
|
+
cert_chain: []
|
12
|
+
date: 2008-01-19 00:00:00.000000000 Z
|
13
|
+
dependencies:
|
14
|
+
- !ruby/object:Gem::Dependency
|
15
15
|
name: stemmer
|
16
|
+
requirement: !ruby/object:Gem::Requirement
|
17
|
+
none: false
|
18
|
+
requirements:
|
19
|
+
- - ! '>='
|
20
|
+
- !ruby/object:Gem::Version
|
21
|
+
version: 1.0.0
|
16
22
|
type: :runtime
|
17
|
-
|
18
|
-
version_requirements: !ruby/object:Gem::Requirement
|
19
|
-
|
20
|
-
|
21
|
-
|
23
|
+
prerelease: false
|
24
|
+
version_requirements: !ruby/object:Gem::Requirement
|
25
|
+
none: false
|
26
|
+
requirements:
|
27
|
+
- - ! '>='
|
28
|
+
- !ruby/object:Gem::Version
|
22
29
|
version: 1.0.0
|
23
|
-
version:
|
24
30
|
description: A general classifier module to allow Bayesian and other types of classifications.
|
25
31
|
email: lucas@rufy.com
|
26
32
|
executables: []
|
27
|
-
|
28
33
|
extensions: []
|
29
|
-
|
30
34
|
extra_rdoc_files: []
|
31
|
-
|
32
|
-
files:
|
35
|
+
files:
|
33
36
|
- lib/classifier.rb
|
34
|
-
- lib/classifier
|
35
37
|
- lib/classifier/bayes.rb
|
36
38
|
- lib/classifier/lsi.rb
|
37
|
-
- lib/classifier/extensions
|
38
39
|
- lib/classifier/extensions/string.rb
|
39
40
|
- lib/classifier/extensions/vector.rb
|
40
41
|
- lib/classifier/extensions/vector_serialize.rb
|
41
42
|
- lib/classifier/extensions/word_hash.rb
|
42
|
-
- lib/classifier/lsi
|
43
43
|
- lib/classifier/lsi/content_node.rb
|
44
44
|
- lib/classifier/lsi/summary.rb
|
45
45
|
- lib/classifier/lsi/word_list.rb
|
46
46
|
- bin/bayes.rb
|
47
47
|
- bin/summarize.rb
|
48
|
-
- test/bayes
|
49
48
|
- test/bayes/bayesian_test.rb
|
50
49
|
- test/test_helper.rb
|
51
|
-
- test/extensions
|
52
50
|
- test/extensions/word_hash_test.rb
|
53
|
-
- test/lsi
|
54
51
|
- test/lsi/lsi_test.rb
|
55
52
|
- README
|
56
53
|
- Rakefile
|
57
54
|
- LICENSE
|
58
|
-
has_rdoc: true
|
59
55
|
homepage: http://classifier.rufy.com/
|
56
|
+
licenses: []
|
60
57
|
post_install_message:
|
61
58
|
rdoc_options: []
|
62
|
-
|
63
|
-
require_paths:
|
59
|
+
require_paths:
|
64
60
|
- lib
|
65
|
-
required_ruby_version: !ruby/object:Gem::Requirement
|
66
|
-
|
67
|
-
|
68
|
-
|
61
|
+
required_ruby_version: !ruby/object:Gem::Requirement
|
62
|
+
none: false
|
63
|
+
requirements:
|
64
|
+
- - ! '>'
|
65
|
+
- !ruby/object:Gem::Version
|
69
66
|
version: 0.0.0
|
70
|
-
|
71
|
-
|
72
|
-
requirements:
|
73
|
-
- -
|
74
|
-
- !ruby/object:Gem::Version
|
75
|
-
version:
|
76
|
-
|
77
|
-
requirements:
|
67
|
+
required_rubygems_version: !ruby/object:Gem::Requirement
|
68
|
+
none: false
|
69
|
+
requirements:
|
70
|
+
- - ! '>='
|
71
|
+
- !ruby/object:Gem::Version
|
72
|
+
version: '0'
|
73
|
+
requirements:
|
78
74
|
- A porter-stemmer module to split word stems.
|
79
75
|
rubyforge_project:
|
80
|
-
rubygems_version: 1.
|
76
|
+
rubygems_version: 1.8.24
|
81
77
|
signing_key:
|
82
78
|
specification_version: 1
|
83
79
|
summary: A general classifier module to allow Bayesian and other types of classifications.
|
84
80
|
test_files: []
|
85
|
-
|