summa 0.0.1 → 0.0.2

data/lib/summa.rb

@@ -2,12 +2,12 @@ $:.unshift(File.dirname(__FILE__)) unless
   $:.include?(File.dirname(__FILE__)) || $:.include?(File.expand_path(File.dirname(__FILE__)))
 
 module Summa
-  VERSION = '0.0.1'
+  VERSION = '0.0.2'
 end
 
 class String
   def stem
-    puts "Just testing!"
+    puts "Still testing!... v0.0.2"
   end
 end
 
@@ -19,70 +19,69 @@ class PheremoneAnalysis
     end
     @document = document
     @keywords = keywords
-    @sigma = constants.sigma
+    @sigma = 16
     @sigma_sq = @sigma * @sigma
-    @threshold = constants.threshold
+    @threshold = 0.8
     @output = ""
   end
 	
-	def analyze()
-		front = 1/(@sigma * Math.sqrt(2 * Math::PI))
-		
-		for i in 0 ... @document.wordArray.length
-			for j in 0 ... @keywords.wordArray.length
-				if @document.docArray[i][@keywords.wordArray[j]] != nil
-					for pos in 0 ... @document.wordArray.length
-						temp = front * Math.exp(-((i-pos)*(i-pos))/(2*@sigma_sq))
-						@pheremones[pos] = @pheremones[pos] + temp
-					end
-				end
-			end
-		end
+  def analyze()
+    front = 1/(@sigma * Math.sqrt(2 * Math::PI))
+    for i in 0 ... @document.wordArray.length
+      for j in 0 ... @keywords.wordArray.length
+        if @document.docArray[i][@keywords.wordArray[j]] != nil
+          for pos in 0 ... @document.wordArray.length
+            temp = front * Math.exp(-((i-pos)*(i-pos))/(2*@sigma_sq))
+	    @pheremones[pos] = @pheremones[pos] + temp
+          end
+        end
+      end
+    end
 		
-		max_value = -1;
-		for i in 0 ... @pheremones.length
-			if max_value < @pheremones[i]
-				max_value = @pheremones[i];
-				end
-			end
+    max_value = -1;
+    for i in 0 ... @pheremones.length
+      if max_value < @pheremones[i]
+        max_value = @pheremones[i];
+      end
+    end
 		
-		for i in 0 ... @pheremones.length
-			@pheremones[i] /= max_value;
-		end
-	end
+    for i in 0 ... @pheremones.length
+      @pheremones[i] /= max_value;
+    end
+  end
 	
-	def summarize()
-		@output = ""
-		inRegion = false
-		for i in 0 ... @pheremones.length
-			if @pheremones[i] >= @threshold
-				if inRegion
-					@output += @document.docArray[i];
-					@output += " ";
-				else
-					inRegion = true
-					@output += findStartOfSentence(i)
-					@output += @document.docArray[i];
-					@output += " "
-				end
-			else
-				if inRegion
-					inRegion = false
-					@output += findEndOfSentence(i-1)
-					@output += " "
-				end
-			end
-		end
-		@output
-	end
+  def summarize()
+    @output = ""
+    inRegion = false
+    for i in 0 ... @pheremones.length
+      if @pheremones[i] >= @threshold
+        if inRegion
+          @output += @document.docArray[i];
+          @output += " ";
+        else
+          inRegion = true
+          @output += findStartOfSentence(i)
+          @output += @document.docArray[i];
+          @output += " "
+        end
+      else
+        if inRegion
+          inRegion = false
+          @output += findEndOfSentence(i-1)
+          @output += " "
+        end
+      end
+    end
+    @output
+  end
 	
-	def findStartOfSentence(i)
-		index = i
-		startIndex = 0
-		local_output = ""
-		while(index > 0)
-			word = @document.docArray[index]
-			if(word != "Mr." && word != "Mrs." && word != "Dr." && 
+  def findStartOfSentence(i)
+    index = i
+    startIndex = 0
+    local_output = ""
+    while(index > 0)
+      word = @document.docArray[index]
+      if(word != "Mr." && word != "Mrs." && word != "Dr." && 
 			   word != "U.S." && word != "Jan." &&
 			   word != "Feb." && word != "Mar." &&
 			   word != "Apr." && word != "May." &&
@@ -92,29 +91,29 @@ class PheremoneAnalysis
 			   word != "Dec." && word != "Sept." &&
 			   word != "Lt." && word != "Maj." &&
 			   word != "Col.")
-				c = word[word.length - 1]
-				if(c == "."[0] || c == ";"[0] || c == ":"[0])
-					startIndex = index + 1
-					break;
-				end
-			end
-			index = index - 1
-		end
+        c = word[word.length - 1]
+        if(c == "."[0] || c == ";"[0] || c == ":"[0])
+          startIndex = index + 1
+          break;
+        end
+      end
+      index = index - 1
+    end
 		
-		for j in startIndex ... i
-			local_output += @document.docArray[j]
-			local_output += " "
-		end
-		local_output
-	end
+    for j in startIndex ... i
+      local_output += @document.docArray[j]
+      local_output += " "
+    end
+    local_output
+  end
 	
-	def findEndOfSentence(i)
-		endIndex = @document.docArray.length
-		index = i
-		local_output = ""
-		while(index < @document.docArray.length)
-			word = @document.docArray[index]
-			if(word != "Mr." && word != "Mrs." && word != "Dr." && 
+  def findEndOfSentence(i)
+    endIndex = @document.docArray.length
+    index = i
+    local_output = ""
+    while(index < @document.docArray.length)
+      word = @document.docArray[index]
+      if(word != "Mr." && word != "Mrs." && word != "Dr." && 
 			   word != "U.S." && word != "Jan." &&
 			   word != "Feb." && word != "Mar." &&
 			   word != "Apr." && word != "May." &&
@@ -124,25 +123,129 @@ class PheremoneAnalysis
 			   word != "Dec." && word != "Sept." &&
 			   word != "Lt." && word != "Maj." &&
 			   word != "Col.")
-				c = word[word.length - 1]
-				if(c == "."[0] || c == ";"[0] || c == ":"[0])
-					endIndex = index
-					break;
+        c = word[word.length - 1]
+        if(c == "."[0] || c == ";"[0] || c == ":"[0])
+            endIndex = index
+            break;
+        end
+      end
+      index = index + 1
+    end
+		
+    if endIndex != i
+      for j in i+1 ... (endIndex + 1)
+        local_output += @document.docArray[j]
+        local_output += " "
+      end
+    end
+		
+    local_output
+  end
+
+  attr_accessor :pheremones, :documet, :keywords, :sigma, :threshold, :output
+end
+
+
+class FrequencyAnalyzer
+	def initialize(document,stopWordsDoc)
+		@freqCount = {}
+		@document = document
+		@stopWords = stopWordsDoc
+		@mean = 0
+		@keywords = [];
+		
+		for i in 0..@document.docArray.length
+			word = @document.docArray[i]
+			if word != nil
+				word = CGWordOps.removePunctuation(word)
+				if !@stopWords.docArray.include?(word.downcase)
+					#stemmed = word.stem
+					if @freqCount.has_key?(word)
+						@freqCount[word] = 
+							@freqCount[word] + 1
+					else
+						@freqCount[word] = 1
+					end
 				end
 			end
-			index = index + 1
 		end
 		
-		if endIndex != i
-			for j in i+1 ... (endIndex + 1)
-				local_output += @document.docArray[j]
-				local_output += " "
+		#@freqCount.each {|key, value| puts "#{key} is #{value}" }
+				
+		sum = 0
+		count = 0
+		keys = @freqCount.keys
+		for i in 0..keys.length
+			if keys[i] != nil
+				sum = sum + @freqCount[keys[i]]
+				count = count + 1
 			end
 		end
+				
+		@mean = sum/count
+	end
+	
+	def analyze(k=3)
+		@keywords = [];
 		
-		local_output
+		keys = @freqCount.keys
+		for i in 0..keys.length
+			if keys[i] != nil
+				value = @freqCount[keys[i]]
+				if value > k * @mean && keys[i] != ""
+					@keywords << keys[i]
+				end
+			end
+		end
+		@keywords
+	end
+	attr_accessor :freqCount, :keywords
+end
+
+
+class CGWordOps
+	def self.removePunctuation(word)
+		output = word;
+		output = output.delete(".")
+		output = output.delete(",")
+		output = output.delete("--")
+		output = output.delete(";")
+		output = output.delete(":")
+		output = output.delete("(")
+		output = output.delete(")")
+		output = output.delete("[")
+		output = output.delete("]")
+		output = output.delete("?")
+		output = output.delete("!")
+		output = output.delete("\"")
+		output
 	end
-	attr_accessor :pheremones, :documet, :keywords, :sigma, :threshold, :output
 end
 
 
+class CGDocument
+	def initialize(docName)  
+		@docName = docName
+		@docFile = File.new(@docName)
+		line=""
+		@docArray = [];
+		@wordArray = [];
+		@docString = "";
+		while line != nil
+			line = @docFile.gets()
+			if line != nil
+				lineArray = line.split(" ")
+				for i in 0 ... lineArray.length
+					@docString += lineArray[i]
+					@docString += " "
+					@docArray << lineArray[i]
+					@wordArray << CGWordOps.removePunctuation(lineArray[i].stem).stem
+				end
+			end
+		end
+	end 
+	#Create getters/setters.
+	attr_accessor :docArray, :wordArray, :docName, :docFile, :docString
+	
+end
+

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: summa
 version: !ruby/object:Gem::Version 
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors: 
 - HyLiter.org
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-09-21 00:00:00 -07:00
+date: 2009-09-22 00:00:00 -07:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency