Research And Implementation Of Intelligent Scoring Simulated Platform For College Entrance Examination Of Chinese Language

The large population and high enthusiasm for taking the college entrance examination in China lead to an extra-high pressure for teachers in the high schools.In addition,the Chinese composition evaluation of examination papers is of greater subjectivity.The educational background and personal preference among the faculty exhibit a high level of subjectivity and volatility to the judgement of objective examination papers.The subject of Chinese language is of great importance to the college entrance examination,which has a great impact on the entrance of candidates.And the influence of subjective writing achievement on candidates cannot be ignored.In order to achieve efficient marking and eliminate the differences between different examiners,it has important practical significance for Chinese teaching and talent selection to develop an exercise and automatic marking system based on computer technology.In our country,there are great differences between Chinese marking and English marking,mainly because of the different characteristics of the two languages.For candidates,the former is their mother tongue and the latter is a foreign language,and their language requirements are different.There are great differences between the two systems in algorithm design and system development.Therefore,although the research on automatic marking of Chinese can draw lessons from the international automatic marking technology and achievements,it is more important to study the characteristics of Chinese itself and the characteristics of Chinese students.Composition marking is the most complicated in the Chinese marking of college entrance examination.The core content of this thesis is to study the marking algorithm of Chinese language composition for Chinese students.The main contents and work of this thesis are as follows:1.Study and analyze the reliability of the composition score prediction system based on content scoring.TextRank and TF-IDF are used to extract and predict the features of composition text.The results show that it is not reliable to judge composition only based on content.2.Proposes the features of writing embedded in language depth perception.Through the analysis of the language sense of Chinese composition in college entrance examination,this thesis puts forward the text characteristics of composition,such as context relevance,fluency of composition sentences,and the use of composition materials,and carries out experiments.Experiments show that it is beneficial to quantify the language sense features and apply them to the automatic scoring of Chinese compositions.3.Research on the realization of marking papers on the simulation platform of college entrance examination.After the analysis of examination requirements,the marking method is divided into three kinds.This thesis regards composition score as a regression problem,and explores the applicability of multiple linear regression,tree regression,support vector regression,gradient lifting decision tree regression and other regression methods to score,which improves the accuracy and reliability of the system.This thesis develops and implements a simulation platform and corresponding marking methods based on Java language and SSH technology.The scoring system of the platform is more suitable for human marking than the automatic marking system based on content theme only.