text2sentiment.rb

def text2sentiment(source, dictionary)
	score = 0;

	# sourceのノーマリゼーション？が必要。全角半角とか。特に数字。
	# 単語が複数回登場することがあるがその場合の評価は？

	dictionary.each{ |dic|
		pattern = dic[:word].gsub(/ /, '.*')	#スペースは単純に削除でも良いかもしれない
		if source.match(/#{pattern}/)
			p "score = " + dic[:score].to_s + ", match = " + pattern
			score += dic[:score]
		end
	}
	score
end

def readDictionary()
	dictionary = []

	#日本語評価極性辞書（用言編）
	begin
		File.open('./dic/wago.121808.pn.txt') do |file|
    	file.each_line do |line|
			label, word = line.chomp.split(/\t/)
			if word.nil? then
				next
			end
			score = if label[0,2] == "ポジ"
				1
			else
				-1
			end
			dictionary.push({:score => score, :word => word})
    	end
	  end
	rescue SystemCallError => e
		puts e.message
	rescue IOError => e
		puts e.message
	end

	#日本語評価極性辞書（名詞編）
	begin
		File.open('./dic/pn.csv.m3.120408.trim') do |file|
    	file.each_line do |line|
			word, label = line.chomp.split(/\t/)
			if word.nil? then
				next
			end
			score = if label == "p"
				1
			elsif label == "n"
				-1
			else
				0
			end
			dictionary.push({:score => score, :word => word})
    	end
	  end
	rescue SystemCallError => e
		puts e.message
	rescue IOError => e
		puts e.message
	end

	dictionary
end



source = "テスト文字列をここに。"

dictionary = readDictionary()

score = text2sentiment(source, dictionary)

p source
p "result = " + score.to_s