Bullet Subtitle Sentiment Classification Based on Affective Computing and Ensemble Learning

Algorithm 2

Input: a bullet subtitle set, the Chinese Bullet Subtitle Sentiment Lexicon

Output: the sentiment category of bullet subtitle

Step 1: split into words and initializing

Step 2: determine whether contains sentiment words, if it does, execute Step3; otherwise, execute Step5,

Step 3: input into BS-CAL described in Algorithm 1 and get the classification result

Step 4: input into Naive Bayes classifier to obtain the classification result

Step 5: use one-hot to process the splited data, and feed into ATT-GRU classifier to get the classification result

Step 6: if , then is the result of the fusion of ; otherwise, is the final result of

Step 7: output the bullet subtitle sentiment category and calculate the accuracy, precision, recall and F1-score.