Research On Automatic Quality Classification Technology Of Chinese Composition In Primary School Based On Multidimensional Features

In Chinese education,how to cultivate and improve students' writing ability has always been the focus.In the traditional teaching process,composition evaluation is mainly based on artificial methods,which has many shortcomings,such as a lot of manpower and material resources are needed,strong subjectivity and long feedback time.With the development of Natural Language Processing technology,the research on the quality of composition at different levels is deepening.At present,there are few studies on Chinese composition in primary school.This study takes Chinese composition of primary school as the research data,uses a variety of artificial intelligence algorithms and combines the development level of writing ability of primary school students,and analyzes and models the composition from many dimensions,such as character,word,sentence,text and theme.The main contents and results of this study are as follows:Firstly,this study extracts 64 features at word level,word level,sentence level and discourse level.Based on these features,the features are screened by random forest,and the grade is used as category,and the support vector machine is used to build the automatic quality classification model of composition.The accuracy rate is 72.73%in low,middle and high grades classification model,and the accuracy rate is 80.86%in low and high grades classification models.Secondly,the thematic features are extracted from the composition data using the LDA theme model in an unsupervised way.Based on the theme features,the support vector mechanism is used to build the classification model.Comparing the second chapter model and the third chapter model,we find that the classification effect is the best under the fusion features.Thirdly,some sentences are scored artificially and used as training data.A sentence scoring model is constructed based on convolution neural network,and the rest sentences are scored by using this model.The quality of continuous sentences reflects the overall quality of the text.Based on sentence scores,this study designs features and construct a classification model.Fourthly,based on the technologies above,an online Chinese composition evaluation prototype system is constructed.The main functions of the system are as follows:calculating and displaying the characteristic data of the composition,recommending the theme similar articles,predicting the grades of the compositions,and scoring the composition sentences automatically.Through this system,students can not only get the detailed analysis data of their compositions,but also browse their historical composition data.In a word,this study applies writing cognition theory,computational linguistics,Natural Language Processing,machine learning and other techniques to analyze Chinese composition deeply.Based on linguistic features,topic characteristics and sentence scoring characteristics,a classification model of primary Chinese composition quality is constructed.This study improves the accuracy of the model by integrating multi-dimensional features and promotes the application and development of the automatic classification technology of composition quality in Chinese teaching.