groupie 0.1.1 → 0.2.2

data/VERSION

@@ -1 +1 @@
-0.1.1
+0.2.2

data/groupie.gemspec

@@ -5,7 +5,7 @@
 
 Gem::Specification.new do |s|
   s.name = %q{groupie}
-  s.version = "0.1.1"
+  s.version = "0.2.2"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Wes Oldenbeuving"]

data/lib/groupie.rb

@@ -12,39 +12,60 @@ class Groupie
     @groups[group] ||= Group.new(group)
   end
 
-  def classify(entry)
+  def classify(entry, strategy=:sum)
     results = {}
     total_count = @groups.inject(0) do |sum, name_group|
       group = name_group.last
-      sum + group.count(entry)
+      count = group.count(entry)
+      if strategy==:sum
+        sum += count
+      elsif strategy==:sqrt
+        sum += Math::sqrt(count)
+      elsif strategy==:log
+        sum += Math::log10(count) if count > 0
+      else
+        raise "Invalid strategy: #{strategy}"
+      end
+      next sum
     end
     return results if 0 == total_count
 
     @groups.each do |name, group|
       count = group.count(entry)
+      if strategy==:sum
+        # keep count
+      elsif strategy==:sqrt
+        count = Math::sqrt(count)
+      elsif strategy==:log
+        count = Math::log10(count) if count > 0
+      else
+        raise "Invalid strategy: #{strategy}"
+      end
       results[name] = count > 0 ? count.to_f / total_count : 0.0
     end
     return results
   end
 
   # Classify a text by taking the average of all word classifications.
-  def classify_text(words)
+  def classify_text(words, strategy=:sum)
+    hits = 0
     group_score_sums = words.inject({}) do |results, word|
-      word_results = classify(word)
+      word_results = classify(word, strategy)
+      next results if word_results.empty?
+      hits += 1
       results.merge(word_results) do |key, old, new|
         old + new
       end
     end
 
-    words_count = words.size.to_f
     averages={}
     group_score_sums.each do |group, sum|
-      averages[group] = sum / words_count
+      averages[group] = hits > 0 ? sum / hits : 0
     end
 
     averages
   end
-  
+
   def self.version
     File.read(File.join(File.dirname(File.expand_path(__FILE__)), "..", "VERSION")).strip
   end

data/spec/groupie_spec.rb

@@ -45,7 +45,37 @@ describe Groupie do
       c2 = g.classify('user')
       c2[:ham].should > c2[:spam]
     end
+
+    describe "strategies" do
+      describe "sum" do
+        it "should weigh words for the sum of their occurances" do
+          g = Groupie.new
+          g[:spam].add %w[word] * 9
+          g[:ham].add %w[word]
+          g.classify('word', :sum).should == {:spam=>0.9, :ham=>0.1}
+        end
+      end
+
+      describe "sqrt" do
+        it "should weigh words for the square root of the sum of ocurances" do
+          g = Groupie.new
+          g[:spam].add %w[word] * 9
+          g[:ham].add %w[word]
+          g.classify('word', :sqrt).should == {:spam=>0.75, :ham=>0.25}
+        end
+      end
+
+      describe "log" do
+        it "should weigh words for log10 of their sum of occurances" do
+          g = Groupie.new
+          g[:spam].add %w[word] * 1000
+          g[:ham].add %w[word] * 10
+          g.classify('word', :log).should == {:spam=>0.75, :ham=>0.25}
+        end
+      end
+    end
   end
+
   context "classify_text" do
     it 'should tokenized html emails' do
       g = Groupie.new
@@ -71,5 +101,30 @@ describe Groupie do
       result2 = g.classify_text "Grow flowers to give to your mom".tokenize
       result2[:ham].should == result2[:spam]
     end
+
+    it "should skip unknown tokens" do
+      g = Groupie.new
+      g[:spam].add %w[buy viagra now]
+      g[:ham].add %w[buy flowers now]
+      g.classify_text(%w[buy buckets now]).should == {:spam=>0.5, :ham=>0.5}
+    end
+    
+    it "should support the sqrt strategy" do
+      g = Groupie.new
+      g[:spam].add %w[one] * 9
+      g[:ham].add %w[one]
+      g[:spam].add %w[two] * 9
+      g[:ham].add %w[two]
+      g.classify_text(%w[one two three], :sqrt).should == {:spam=>0.75, :ham=>0.25}
+    end
+    
+    it "should support the log strategy" do
+      g = Groupie.new
+      g[:spam].add %w[one] * 100
+      g[:ham].add %w[one]
+      g[:spam].add %w[two]
+      g[:ham].add %w[two] * 100
+      g.classify_text(%w[one two three], :log).should == {:spam=>0.5, :ham=>0.5}
+    end
   end
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: groupie
 version: !ruby/object:Gem::Version 
-  hash: 25
+  hash: 19
   prerelease: false
   segments: 
   - 0
-  - 1
-  - 1
-  version: 0.1.1
+  - 2
+  - 2
+  version: 0.2.2
 platform: ruby
 authors: 
 - Wes Oldenbeuving