bayes_on_redis 0.1.0

data/README.markdown

@@ -0,0 +1,13 @@
+# bayes_on_redis
+
+Bayesian classifier on top of Redis
+
+## Why on Redis?
+
+Redis is perfect for building fast bayesian filter.
+
+## Getting started
+
+## Contributing
+
+[Fork the project](http://github.com/didip/bayes_on_redis) and send pull requests.

data/lib/bayes_on_redis.rb

@@ -0,0 +1,74 @@
+require "rubygems"
+require "redis"
+
+class BayesOnRedis
+  CATEGORIES_KEY = "BayesOnRedis:categories"
+
+  def initialize(options)
+    @redis = Redis.new(:host => options[:redis_host], :port => options[:redis_port], :db => options[:redis_db])
+  end
+
+  def flushdb
+    @redis.flushdb
+  end
+
+  # training for a category
+  def train(category, text)
+    category = category.downcase
+    @redis.sadd(CATEGORIES_KEY, category)
+
+    count_occurance(text).each do |word, count|
+      @redis.hincrby(redis_category_key(category), word, count)
+    end
+  end
+
+  def untrain(category, text)
+    category = category.downcase
+
+    count_occurance(text).each do |word, count|
+      word_count_atm = @redis.hget(redis_category_key(category), word)
+      if (word_count_atm >= count)
+        new_count = (word_count_atm - count)
+      else
+        new_count = 0
+      end
+      @redis.hset(redis_category_key(category), word, new_count)
+    end
+  end
+
+  def classify(text)
+    scores = {}
+
+    @redis.smembers(CATEGORIES_KEY).each do |category|
+      words_count_per_category = @redis.hvals(redis_category_key(category)).inject(0) {|sum, score| sum + score.to_i}
+      @redis.srem(CATEGORIES_KEY, category) if words_count_per_category <= 0
+
+      scores[category] = 0
+
+      count_occurance(text).each do |word, count|
+        tmp_score = @redis.hget(redis_category_key(category), word).to_i
+        tmp_score = 0.1 if tmp_score <= 0
+
+        scores[category] += Math.log(tmp_score / words_count_per_category.to_f)
+      end
+    end
+
+    return scores
+  end
+
+  def classify_for_human(text)
+    (classify(text).sort_by { |score| -score[1] })[0][0]    # [0][0] -> first score, get the key
+  end
+
+  private
+  def redis_category_key(category)
+    "BayesOnRedis:cat:#{category}"
+  end
+
+  # Incoming text is always downcased
+  def count_occurance(text)
+    text.downcase.split.inject(Hash.new(0)) do |container, word|
+      container[word] += 1; container
+    end
+  end
+end

metadata

@@ -0,0 +1,68 @@
+--- !ruby/object:Gem::Specification 
+name: bayes_on_redis
+version: !ruby/object:Gem::Version 
+  hash: 27
+  prerelease: false
+  segments: 
+  - 0
+  - 1
+  - 0
+  version: 0.1.0
+platform: ruby
+authors: 
+- Didip Kerabat
+autorequire: 
+bindir: bin
+cert_chain: []
+
+date: 2010-11-24 00:00:00 -08:00
+default_executable: 
+dependencies: []
+
+description: bayes_on_redis library provides bayesian classification on a given text similar to many SPAM/HAM filtering.
+email: didipk@gmail.com
+executables: []
+
+extensions: []
+
+extra_rdoc_files: []
+
+files: 
+- README.markdown
+- lib/bayes_on_redis.rb
+has_rdoc: true
+homepage: https://github.com/didip/bayes_on_redis
+licenses: []
+
+post_install_message: 
+rdoc_options: []
+
+require_paths: 
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement 
+  none: false
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      hash: 3
+      segments: 
+      - 0
+      version: "0"
+required_rubygems_version: !ruby/object:Gem::Requirement 
+  none: false
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      hash: 3
+      segments: 
+      - 0
+      version: "0"
+requirements: []
+
+rubyforge_project: 
+rubygems_version: 1.3.7
+signing_key: 
+specification_version: 3
+summary: Bayesian filter on top of Redis
+test_files: []
+