Formal and Informal detection in textual data

I need to find out the formality and informality in textual data. Things you have to do: 1. Suggest the possible approaches (more than one) through which the formality of text can be judged. 2. Implement one of the approaches which you think would be the best fit for the problem. Note: If the expected time and resources for training are high then you can reduce the no. of epochs as well as the no. of training examples. 3. Analyze the results and point out the possible reasons for the false cases. About Data: There are two data files labels_train, labels_test consisting of train and test data respectively. The structure of the files is example, score where the score is from -3.0 to 3.0, -3.0 being very informal and 3.0 very formal