Ensemble Learning on Scoring Student Essay

摘要

Automated essay scoring is becoming more and more concerned by the researchers. In this work, we develop a new way to extract Textual features, which is proved to be valid. First, we calculate the Distributed Representation from the WiKi corpus by the word2vec. Then we calculate the number of words, the number of dictionary, the diversity of words as the textual features by K-means and Distributed Representation. There will be 3*k textual features as the k represents the number of categories. Besides, we calculate the structure features including the length of essay, the number of paragraph, the length of sentence etc. We use several models such as XGBoost, Random Forest, GBDT to train the training set and predict the test set. Finally, We ensemble the prediction of those models as the final prediction.