Predicting Reading Comprehension Score for Elementry Students

Usage

• BiLSTM.py contains the bidirecional LSTM model designed for the prediction task.
• train.py includes all functions to train, evaluate, predict with the BiLSTM Model
• generate_dataset.py reads the raw data and generate dataset with "text" and "score" for the prediction task.
• split_dataset.py is a utility function to split the dataset into training and testing set with a given percentage.
• tokenize_word_embed.py is a testing script for developing tokenizers and check embedding coverage on tokenized text.
• create_embed.py generates a embedding pickle file that indexes vocab of a given corpus and appends GloVe embedding to each word.