from count_smilies_class import Emo from collections import Counter import os import json import codecs import numpy as np numOfTotalSentences = 0 messagecounter = 0 foundEmosAndSmilies = [] # emoCount = Emo(language="empty", emoticons=True, secondemo="emoticons") emoCount = Emo(language="twitchstandard", emoticons=True, secondemo="emoticons") # import messages and parse def getData(filename="", dirname='txtfiles/messages'): messages = [] if filename == "": for fname in os.listdir(dirname): print os.listdir(dirname), fname for line in open(os.path.join(dirname, fname)): try: parsed = json.loads(line) parsed_msg = parsed['msg'] messages.append(parsed_msg) except ValueError: print "Error in line" continue else: for line in open(os.path.join(dirname, filename)):
# if score > 0: # score = float(1.0) # elif score < 0: # score = float(-1.0) # print labels[index], score # if labels[index] == score: # rightSentimentCount += 1 # # predictedlabels.append(score) """ Smilie scoring """ from count_smilies_class import Emo emoCount = Emo(language="twitchstandard", emoticons=False, secondemo="emoticons") for index, sent in enumerate(messages): sentenceCount += 1 # print sentenceCount print sent score, words = emoCount.score(sent) # bisschen umgeschrieben, dass auch die missing words ausgegeben werden a = emoCount.find_all(sent) if len(a) == 0: emptysents += 1 # print score if score > 0: score = float(1.0) elif score < 0: score = float(-1.0) print labels[index], score if labels[index] == score: