judgee 0.1.0

data/Gemfile

@@ -0,0 +1,4 @@
+source 'https://rubygems.org'
+
+# Specify your gem's dependencies in judgee.gemspec
+gemspec

data/LICENSE.txt

@@ -0,0 +1,22 @@
+Copyright (c) 2013 Railsmechanic
+
+MIT License
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
+LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

data/README.md

@@ -0,0 +1,54 @@
+# Judgee
+
+A simple Bayesian Classifier with additive smoothing built in.
+The primary focus of judgee lies on performance and a minimal but flexible feature set.
+So it's up to you to do stemming, text analysis, etc.
+
+
+## Backed by Redis
+
+[Redis](http://redis.io/) is an open source, BSD licensed, advanced key-value store, which is often referred to as a data structure server.
+It supports strings, hashes, lists, sets, sorted sets and offers an incredible performance.
+
+
+## Installation
+
+	gem install judgee
+
+
+## Getting started
+
+	# Require Judgee
+	require "judgee"
+  
+	# Create an instance of Judgee.
+	# It assumes that your Redis instance is running on localhost at port 6379.
+	judgee = Judgee::Classifier.new
+
+	# Is Redis running on a host in your network, simply pass in your options
+	judgee = Judgee::Classifier.new(:host => "10.0.1.1", :port => 6380)
+
+	# It also supports Unix sockets
+	judgee = Judgee::Classifier.new(:path => "/tmp/redis.sock")
+
+
+
+	# Now you can train the classifier
+	judgee.train(:spam, ["bad", "worse", "stupid", "idiotic"])
+	judgee.train(:ham, ["good", "better", "best", "lovely"])
+
+	# After training, classify your text sample
+	judgee.classify(["good", "better", "best", "worse"]) # => :ham
+
+
+	# Want to untrain some words?
+	judgee.untrain(:spam, ["bad", "worse"])
+
+
+## Information on Performance 
+
+If you read the source code, you might stumble upon the confusing method names.
+There are two methods for training (train, train_fast), two methods for untraining (untrain, untrain_fast) and two methods for classification (classify, classify_fast).
+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.
+
+So use the '_fast' methods if you need performance, e.g. in production and the methods without the suffix for learning purposes.

data/Rakefile

@@ -0,0 +1 @@
+require "bundler/gem_tasks"

data/judgee.gemspec

@@ -0,0 +1,29 @@
+# -*- encoding: utf-8 -*-
+
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'judgee/version'
+
+Gem::Specification.new do |gem|
+  gem.name          = "judgee"
+  gem.version       = Judgee::VERSION
+  gem.authors       = ["Railsmechanic"]
+  gem.email         = ["info@railsmechanic.de"]
+  gem.description   = %q{A simple Bayesian Classifier with additive smoothing and its focus on performance.}
+  gem.summary       = %q{Judgee is a simple Bayesian Classifier with additive smoothing, which uses Redis for persistance.}
+  gem.homepage      = "https://github.com/railsmechanic/judgee"
+  gem.homepage      = "https://github.com/railsmechanic/judgee"
+
+  gem.files         = `git ls-files`.split($/)
+  gem.executables   = gem.files.grep(%r{^bin/}).map{ |f| File.basename(f) }
+  gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
+  gem.require_paths = ["lib"]
+  
+  
+  # Regular dependencies
+  gem.add_dependency "redis"
+  
+  # Development dependencies
+  gem.add_development_dependency "rspec"
+  
+end

data/lib/judgee.rb

@@ -0,0 +1,3 @@
+
+require "judgee/version"
+require "judgee/classifier"

data/lib/judgee/classifier.rb

@@ -0,0 +1,138 @@
+# encoding: UTF-8
+
+require "redis"
+
+module Judgee
+  class Classifier
+
+    ###
+    # Constants
+    ###
+    CATEGORIES_KEY  = "judgee:categories"
+    CATEGORY_KEY    = "judgee:category"
+    ALPHA           = 1.0
+
+
+    attr_reader :redis
+
+    def initialize(options={})
+      @redis = Redis.new(options)
+    end
+
+
+    def train(category, data)
+      redis.sadd(CATEGORIES_KEY, category_name(category))
+      count_occurance(data).each do |word, word_count|
+        redis.hincrby(redis_category_key(category), word, word_count)
+      end
+      "OK"
+    end
+
+    def train_fast(category, data)
+      redis.sadd(CATEGORIES_KEY, category_name(category))
+      occurances          = count_occurance(data)
+      database_occurances = Hash[occurances.keys.zip(redis.hmget(redis_category_key(category), occurances.keys))]
+      new_occurances      = occurances.merge(database_occurances) { |key, value_occurance, value_database_occurance| value_occurance.to_i + value_database_occurance.to_i }.to_a.flatten!
+      redis.hmset(redis_category_key(category), new_occurances)
+      "OK"
+    end
+
+
+
+    def untrain(category, data)
+      count_occurance(data).each do |word, word_count|
+        new_count = [(redis.hget(redis_category_key(category), word).to_i - word_count), 0].max
+        if new_count > 0
+          redis.hset(redis_category_key(category), word, new_count)
+        else
+          redis.hdel(redis_category_key(category), word)
+        end
+      end
+      "OK"
+    end
+
+    def untrain_fast(category, data)
+      occurances          = count_occurance(data)
+      database_occurances = Hash[occurances.keys.zip(redis.hmget(redis_category_key(category), occurances.keys))]
+      untrain_occurances  = database_occurances.merge(occurances) { |key, value_occurance, value_untrain_occurance| value_occurance.to_i - value_untrain_occurance.to_i }
+      empty_occurances    = untrain_occurances.select { |key, value| value.to_i <= 0 }
+      redis.hmset(redis_category_key(category), untrain_occurances.to_a.flatten!)
+      redis.hdel(redis_category_key(category), empty_occurances.keys) unless empty_occurances.empty?
+      "OK"
+    end
+
+
+
+    def classify(data)
+      result      = Hash.new(0)
+      categories  = redis.smembers(CATEGORIES_KEY)
+
+      categories.each do |category|
+        count_occurance(data).each do |word, word_count|
+          numerator   = (redis.hget(redis_category_key(category), word).to_i + ALPHA).to_f
+          denominator = (categories.map { |category| redis.hget(redis_category_key(category), word).to_i }.inject(0, :+) + (ALPHA * data.length)).to_f
+          result[category] += (word_count * Math.log(numerator / denominator)).abs
+        end
+      end
+
+      result.min_by(&:last).first.to_sym
+    end
+
+    def classify_fast(data)
+      result      = Hash.new(0)
+      categories  = redis.smembers(CATEGORIES_KEY)
+      occurances  = count_occurance(data)
+
+      categories.each do |category|
+        numerator   = Hash[occurances.keys.zip(redis.hmget(redis_category_key(category), occurances.keys))].inject({}) { |hash, (key, value)| hash[key] = value.to_f + ALPHA; hash }
+        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 }
+        result[category] += numerator.merge(denominator) { |key, value_numerator, value_denominator| (occurances[key] * Math.log(value_numerator / value_denominator)).abs }.values.inject(0, :+)
+      end
+
+      result.min_by(&:last).first.to_sym
+    end
+
+
+
+    def flushdb(flush_db=false)
+      redis.flushdb if flush_db
+    end
+
+
+
+    def flush_category(category)
+      redis.del(redis_category_key(category))
+      redis.srem(CATEGORIES_KEY, category_name(category))
+    end
+
+
+    private
+
+
+    def count_occurance(data='')
+      bag_of_words = Hash.new(0)
+
+      data = [data].flatten.map! do |word|
+        word.to_s.strip
+      end.delete_if(&:empty?)
+
+      for word in data
+        bag_of_words[word] += 1
+      end
+      bag_of_words
+    rescue
+      raise ArgumentError, 'input must be a single String or an Array of Strings'
+    end
+
+
+    def category_name(category)
+      category.to_s.downcase.strip.gsub(/[\s\W]+/,'_').gsub(/_+$/,'')
+    end
+
+
+    def redis_category_key(category)
+      "#{CATEGORY_KEY}:#{category_name(category)}"
+    end
+
+  end
+end

data/lib/judgee/version.rb

@@ -0,0 +1,5 @@
+# encoding: UTF-8
+
+module Judgee
+  VERSION = "0.1.0"
+end

data/spec/judgee_spec.rb

@@ -0,0 +1,36 @@
+# encoding: UTF-8
+
+require 'judgee'
+require 'redis'
+
+describe Judgee::Classifier do  
+  
+  CATEGORIES_KEY  = "judgee:categories"
+  CATEGORY_KEY    = "judgee:category"
+  
+  before :each do
+    @judgee         = Judgee::Classifier.new
+    @redis          = Redis.new
+    @redis.flushdb
+    @spam_category  = :spam_spec
+    @ham_category   = :ham_spec
+    @spam_data      = %w(money rich quick big viagra penis)
+    @ham_data       = %w(mail google gmail maps ruby)
+    @judgee.flush_category(@spam_category)
+    @judgee.flush_category(@ham_category)
+  end
+  
+  describe "training" do
+    it "should add category to 'judgee:categories' set" do
+      @judgee.train(@ham_category, @ham_data)
+      @judgee.train(@spam_category, @spam_data)
+      categories = @redis.smembers(CATEGORIES_KEY)
+      categories.length.should eq 2
+      categories.should_include("judgee:categories:spam_spec")
+      categories.should_include("judgee:categories:ham_spec")
+    end
+  end
+  
+  # TODO
+  
+end

metadata

@@ -0,0 +1,89 @@
+--- !ruby/object:Gem::Specification
+name: judgee
+version: !ruby/object:Gem::Version
+  version: 0.1.0
+  prerelease: 
+platform: ruby
+authors:
+- Railsmechanic
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2013-03-23 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: redis
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: rspec
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: '0'
+description: A simple Bayesian Classifier with additive smoothing and its focus on
+  performance.
+email:
+- info@railsmechanic.de
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- Gemfile
+- LICENSE.txt
+- README.md
+- Rakefile
+- judgee.gemspec
+- lib/judgee.rb
+- lib/judgee/classifier.rb
+- lib/judgee/version.rb
+- spec/judgee_spec.rb
+homepage: https://github.com/railsmechanic/judgee
+licenses: []
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubyforge_project: 
+rubygems_version: 1.8.24
+signing_key: 
+specification_version: 3
+summary: Judgee is a simple Bayesian Classifier with additive smoothing, which uses
+  Redis for persistance.
+test_files:
+- spec/judgee_spec.rb