Design And Development Of A Competition System Supporting Chinese Essay Auxiliary Scoring

In the student-centered educational model reform,it is an innovative measure for students to participate in various subject competitions.However,the offline holding process of subject competitions is complicated and the management work is cumbersome,and the current systems supporting their online holdings generally have problems such as only offering simple functions and lacking of scoring support.Especially for essay competitions,the judges have a large workload and work inefficiently.In order to solve the above problems,this thesis has done the following work:In view of the difficulty of scoring in essay writing competitions,this thesis has investigated the research on essay auxiliary scoring technology,and innovatively propose an auxiliary scoring scheme from different dimensions,including essay off-topic detection and automated essay scoring methods:(1)Starting from the evaluation dimension of consistency between essay and topic,an offtopic detection method based on deep learning is proposed.The method combines bidirectional long short-term memory network and multi-head attention mechanism,and judges whether the composition is off-topic by comparing the Euclidean distance between the vector representation of the topic and the vector representation of the essay.(2)Synthesizing the evaluation dimensions of semantic expression,syntactic structure and vocabulary,an automated essay scoring method is proposed.The method calculates the specific score of an essay by combining the semantic feature vector,syntactic feature vector and lexical feature vector of the essay.This thesis has collected 7284 essay answer sheets of the Hanyu Shuiping Kaoshi(HSK)as the experimental data set,and has tested the above two methods respectively.The experimental results show that the F1 value of the off-topic detection method reaches 0.922,and has made 11.62% improvements over multiple baseline models;the QWK value of the automated essay scoring method reaches 0.711,and has made 7.4% improvements over multiple baseline models.This thesis has investigated the current research on the holding process of the subject competition and the competition system and has designed and implemented a subject competition system.The system is divided into two modules according to functional areas.The system management module is responsible for the daily maintenance and management of the system,and the competition module provides competition-related functional services for the competition administrators,competition team members and competition judges.The system includes an auxiliary scoring module,which realizes the application of off-topic detection and automatic essay scoring methods.The competition system implemented in this thesis has been applied to several large-scale essay competitions and small-scale course seminars to demonstrate the functional integrity and practicability of the system.In addition,this thesis simulates a real scene,and uses the auxiliary scoring module to grade collections of essays ranging from 10 to 200,and counts the time spent grading,and obtains that the average scoring time of each essay is 0.2 seconds,which means the auxiliary scoring module has achieved certain practical effects.