| Scoring in teaching process has an important position. It’s one of a useful tool for student to test their study ability. Also, it’s a great method for the regulator to judge educational level. And the key of scoring is fairness and. With the developing of artificial intelligence(AI), automated grading system comes out, and helps people to grade automatically. Automated essay scoring has objective, neutral, high efficiency and other excellent characteristics that manual scoring do not have.National) College Entrance Examination(NCEE) is one of the most important exams in China. Every year, more than one million students enter NCEE and try their best to get their qualification to enter the college or university.As we see, the Chinese writing part of NCEE is one of the application scenarios of scoring. Examines of NCEE have high language ability, comprehension ability, logical thinking and literature accomplishment. So, it’s really hard for normal people to scoring the essay. On the other hand scoring Chinese essay is time-consuming, but time limited.So, it is both challenge and opportunity for us.In this paper, we have done some interesting work about automated essay scoring below.First, we proposed some novel evaluation methods for our task. It is a foundation work. Before all of these studies, we must abstract the key process of NCEE Chinese writing scoring.Second, we built a regression model based on simple language features, and treat this result as our baseline. We also summarize a scientific way to find features and train the model.Third, we built a regression model based on deep language features to improve our result. We conclude all of our work above. So that we find out the key part, then focus on it.At last, we try some deep learning method. Transfer paragraph to vector then train a model for our task. We try to solve the most difficult part in our task, so we want to automatically find our features to model a accurate regression system of scoring. |
PDF Full Text Request