# RB: Ruble-based, LB: Lexicon-base, ML: Machine Learning classifier count = {'RB': 0, 'LB': 0, 'ML': 0} # Evaluate if tested with the gold standard guess = list() gold = list() # Keep the predictions string output = '' # Load test set tweets #tweets = [tweet['MESSAGE'] for tweet in devset] tweets = [tweet['MESSAGE'] for tweet in testset] # Classify each instance in the testeset in the TwitterHybridClassifier loaded before predictions = classifier.classify_batch(tweets) print("features test size -> " + str(len(var.features_test))) #import pickle #pickle_out = open("../../featuresTest.pickle","ab") #pickle.dump(var.features_test, pickle_out) #pickle_out.close() # Output the semeval prediction file and the evaluation variables # if testset if provided if len(testset) > 0: for index, tweet in enumerate(testset): prediction, method = predictions[index] count[method] += 1
# RB: Ruble-based, LB: Lexicon-base, ML: Machine Learning classifier count = {'RB':0, 'LB':0, 'ML':0 } # Evaluate if tested with the gold standard guess = list() gold = list() # Keep the predictions string output = '' # Load test set tweets #tweets = [tweet['MESSAGE'] for tweet in devset] tweets = [tweet['MESSAGE'] for tweet in testset] # Classify each instance in the testeset in the TwitterHybridClassifier loaded before predictions = classifier.classify_batch(tweets) # Output the semeval prediction file and the evaluation variables # if testset if provided if len(testset) > 0: for index, tweet in enumerate(testset): prediction,method = predictions[index] count[method] += 1 output += tweet['SID'] + '\t' + tweet['UID'] + '\t' + prediction + '\t' + tweet['MESSAGE'] + '\n' guess.append(prediction) gold.append(tweet['SENTIMENT']) confusion_matrix(gold,guess) # Write Semeval output file output_file = 'task9-NILC_USP-B-twitter-constrained.output' codecs.open(output_file,'w','utf8').write(output)