| The automated essay scoring is that the computer can evaluate and score the essay.With the great progress of natural language processing,it becomes possible to evaluate and score Chinese essay by computer. Essay must be tested in a variety of Chinese exams, and there are so many Chinese student take part in college entrance examination and senior high school entrance examination,and there are more and more foreign students take part in HSK in recent years.The automated essay scoring is highly effective and objective,so it is necessary to research on automated Chinese essay scoring.Chinese essay is an examination by national students, it is different from foreign automated essay scoring,Chinese and English writing are very different too,so we can ‘t simply copy other’s practice.We proposed the evaluation method of Chinese essay system, and we conducted in-depth research in the two aspects of the essay literary and composition topic.we also use deep learning technology to automatically learn the feature of the composition.1 We use parallelism and metaphor to represent the feature of the essay literary,and summarize the types of parallelism using.We propose a heuristic method to identify parallelism and metaphor.Using Trie data structure to store ancient poetry resources,it helps us to retrieve the ancient poetry that appeared in the essay.Adding these features to the baseline system,the performance of the system has been improved.2 We use statistical and Text Rank method to automatically extract key words of the essay.We also use LDA automatically learn the topics of the essays.The method of automatically evaluating the quality of each topic has been proposed.Add these topic features to the baseline system,the performance of the system also has been improved.3 Using restricted Boltzmann machine learning vector representation of the essay, this method is helpful for improving the recall rate of the first class of the essay. |
PDF Full Text Request