Emotion classification
Step 1: evaluation
We used precision, recall, accuracy and F1 Measure as our evaluation criterion. dev.csv stores our gold data, dev-predicted.csv stores our predicted data. We got 8 kinds of emotion from the predicated data. We stores the emotion in a list which is emotion_label. We implemented an evaluate function in Class Corpus. The function could calculate the value of the evaluation criterion.
Step 2: data preprocessing
We are now working the data preprocessing to make the data more clean which is used for the input of the classification function.