judgee 0.1.0 → 0.1.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +15 -0
- data/README.md +14 -2
- data/lib/judgee/classifier.rb +5 -7
- data/lib/judgee/version.rb +1 -1
- metadata +4 -10
checksums.yaml
ADDED
@@ -0,0 +1,15 @@
|
|
1
|
+
---
|
2
|
+
!binary "U0hBMQ==":
|
3
|
+
metadata.gz: !binary |-
|
4
|
+
Yjg5MGQ0YjhlNTcwYmM4ZTMwNjgxMzBkMGEyNjNmOWI4NWNhZDJmNg==
|
5
|
+
data.tar.gz: !binary |-
|
6
|
+
ZTkwY2E4ZjZiZjUxNzljOWQwNTcwOWRjZTFiNmVmYmQyODVkNGFjZQ==
|
7
|
+
!binary "U0hBNTEy":
|
8
|
+
metadata.gz: !binary |-
|
9
|
+
MjJlYTFhMmQ4MzVmZDEyMjk1ZjdlYjk0MjdlOWU3YzI0NzI3ODE4YWIyY2U0
|
10
|
+
NWJjOTE2ZmQ2ZjMwMjg5NTJjODUxZmU3Yzg4ODRiZDdkZTM5ZGQyY2Q3ZTFj
|
11
|
+
NTJiMmUzOWE3YTMzZWU1NGRlOTljM2MzZGI1NmEyMWRlNTQzOTA=
|
12
|
+
data.tar.gz: !binary |-
|
13
|
+
ZWQwZmJlNDkxOTgxNmJlNWFjYTM0MWE2Mjc1NzYyMWNlYzZlMTZkNTQxM2Nl
|
14
|
+
ODUxYTkzZjU1OWRmNWIzNGQ5OWJhOThmYzMzOTg5NTU3NDdiYTQ2ZTNlNDMx
|
15
|
+
MGMyMWE0MWNhMTQyZmNlYzgwODFmYmVlYTMxODhhYTkyMGY5ZjU=
|
data/README.md
CHANGED
@@ -32,7 +32,6 @@ It supports strings, hashes, lists, sets, sorted sets and offers an incredible p
|
|
32
32
|
judgee = Judgee::Classifier.new(:path => "/tmp/redis.sock")
|
33
33
|
|
34
34
|
|
35
|
-
|
36
35
|
# Now you can train the classifier
|
37
36
|
judgee.train(:spam, ["bad", "worse", "stupid", "idiotic"])
|
38
37
|
judgee.train(:ham, ["good", "better", "best", "lovely"])
|
@@ -51,4 +50,17 @@ If you read the source code, you might stumble upon the confusing method names.
|
|
51
50
|
There are two methods for training (train, train_fast), two methods for untraining (untrain, untrain_fast) and two methods for classification (classify, classify_fast).
|
52
51
|
The difference is quite simple. As the name suggests, all methods with the suffix '_fast' are (really) faster (3x to 10x) in processing the data, but virtually unreadable.
|
53
52
|
|
54
|
-
So use the '_fast' methods if you need performance, e.g. in production and the methods without the suffix for learning purposes.
|
53
|
+
So use the '_fast' methods if you need performance, e.g. in production and the methods without the suffix for learning purposes.
|
54
|
+
|
55
|
+
## For higher performance, use the _fast methods
|
56
|
+
|
57
|
+
# Now you can train the classifier
|
58
|
+
judgee.train_fast(:spam, ["bad", "worse", "stupid", "idiotic"])
|
59
|
+
judgee.train_fast(:ham, ["good", "better", "best", "lovely"])
|
60
|
+
|
61
|
+
# After training, classify your text sample
|
62
|
+
judgee.classify_fast(["good", "better", "best", "worse"]) # => :ham
|
63
|
+
|
64
|
+
|
65
|
+
# Want to untrain some words?
|
66
|
+
judgee.untrain_fast(:spam, ["bad", "worse"])
|
data/lib/judgee/classifier.rb
CHANGED
@@ -5,9 +5,7 @@ require "redis"
|
|
5
5
|
module Judgee
|
6
6
|
class Classifier
|
7
7
|
|
8
|
-
###
|
9
|
-
# Constants
|
10
|
-
###
|
8
|
+
### Constants ###
|
11
9
|
CATEGORIES_KEY = "judgee:categories"
|
12
10
|
CATEGORY_KEY = "judgee:category"
|
13
11
|
ALPHA = 1.0
|
@@ -84,8 +82,8 @@ module Judgee
|
|
84
82
|
occurances = count_occurance(data)
|
85
83
|
|
86
84
|
categories.each do |category|
|
87
|
-
numerator
|
88
|
-
denominator
|
85
|
+
numerator = Hash[occurances.keys.zip(redis.hmget(redis_category_key(category), occurances.keys))].inject({}) { |hash, (key, value)| hash[key] = value.to_f + ALPHA; hash }
|
86
|
+
denominator = categories.map { |category| Hash[occurances.keys.zip(redis.hmget(redis_category_key(category), occurances.keys))] }.inject(Hash.new(0)) { |main_hash, sub_hash| main_hash.merge(sub_hash) { |key, value_first, value_second| value_first.to_f + value_second.to_f} }.inject(Hash.new(0)) { |hash, (key, value)| hash[key] = value.to_f + (ALPHA * data.length); hash }
|
89
87
|
result[category] += numerator.merge(denominator) { |key, value_numerator, value_denominator| (occurances[key] * Math.log(value_numerator / value_denominator)).abs }.values.inject(0, :+)
|
90
88
|
end
|
91
89
|
|
@@ -112,7 +110,7 @@ module Judgee
|
|
112
110
|
def count_occurance(data='')
|
113
111
|
bag_of_words = Hash.new(0)
|
114
112
|
|
115
|
-
data = [data].flatten.map
|
113
|
+
data = [data].flatten.map do |word|
|
116
114
|
word.to_s.strip
|
117
115
|
end.delete_if(&:empty?)
|
118
116
|
|
@@ -121,7 +119,7 @@ module Judgee
|
|
121
119
|
end
|
122
120
|
bag_of_words
|
123
121
|
rescue
|
124
|
-
raise ArgumentError, '
|
122
|
+
raise ArgumentError, 'Input must be a single String or an Array of Strings'
|
125
123
|
end
|
126
124
|
|
127
125
|
|
data/lib/judgee/version.rb
CHANGED
metadata
CHANGED
@@ -1,8 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: judgee
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.1.
|
5
|
-
prerelease:
|
4
|
+
version: 0.1.1
|
6
5
|
platform: ruby
|
7
6
|
authors:
|
8
7
|
- Railsmechanic
|
@@ -14,7 +13,6 @@ dependencies:
|
|
14
13
|
- !ruby/object:Gem::Dependency
|
15
14
|
name: redis
|
16
15
|
requirement: !ruby/object:Gem::Requirement
|
17
|
-
none: false
|
18
16
|
requirements:
|
19
17
|
- - ! '>='
|
20
18
|
- !ruby/object:Gem::Version
|
@@ -22,7 +20,6 @@ dependencies:
|
|
22
20
|
type: :runtime
|
23
21
|
prerelease: false
|
24
22
|
version_requirements: !ruby/object:Gem::Requirement
|
25
|
-
none: false
|
26
23
|
requirements:
|
27
24
|
- - ! '>='
|
28
25
|
- !ruby/object:Gem::Version
|
@@ -30,7 +27,6 @@ dependencies:
|
|
30
27
|
- !ruby/object:Gem::Dependency
|
31
28
|
name: rspec
|
32
29
|
requirement: !ruby/object:Gem::Requirement
|
33
|
-
none: false
|
34
30
|
requirements:
|
35
31
|
- - ! '>='
|
36
32
|
- !ruby/object:Gem::Version
|
@@ -38,7 +34,6 @@ dependencies:
|
|
38
34
|
type: :development
|
39
35
|
prerelease: false
|
40
36
|
version_requirements: !ruby/object:Gem::Requirement
|
41
|
-
none: false
|
42
37
|
requirements:
|
43
38
|
- - ! '>='
|
44
39
|
- !ruby/object:Gem::Version
|
@@ -62,27 +57,26 @@ files:
|
|
62
57
|
- spec/judgee_spec.rb
|
63
58
|
homepage: https://github.com/railsmechanic/judgee
|
64
59
|
licenses: []
|
60
|
+
metadata: {}
|
65
61
|
post_install_message:
|
66
62
|
rdoc_options: []
|
67
63
|
require_paths:
|
68
64
|
- lib
|
69
65
|
required_ruby_version: !ruby/object:Gem::Requirement
|
70
|
-
none: false
|
71
66
|
requirements:
|
72
67
|
- - ! '>='
|
73
68
|
- !ruby/object:Gem::Version
|
74
69
|
version: '0'
|
75
70
|
required_rubygems_version: !ruby/object:Gem::Requirement
|
76
|
-
none: false
|
77
71
|
requirements:
|
78
72
|
- - ! '>='
|
79
73
|
- !ruby/object:Gem::Version
|
80
74
|
version: '0'
|
81
75
|
requirements: []
|
82
76
|
rubyforge_project:
|
83
|
-
rubygems_version:
|
77
|
+
rubygems_version: 2.0.3
|
84
78
|
signing_key:
|
85
|
-
specification_version:
|
79
|
+
specification_version: 4
|
86
80
|
summary: Judgee is a simple Bayesian Classifier with additive smoothing, which uses
|
87
81
|
Redis for persistance.
|
88
82
|
test_files:
|