Research On The Application Of Multi-dimensional Course Evaluation Model Based On Sentiment Analysis

With the vigorous development of national education,the society’s attention to the quality of teaching in colleges and universities is also increasing.Student teaching evaluation is a key part of the current teaching quality evaluation work.How to deeply mine and use the student teaching evaluation data to improve the teaching quality of colleges and universities has become the focus of many researchers at present.There are two main types of teaching evaluation data in colleges and universities,namely structured teaching scoring data and unstructured course evaluation text data.But at present,most of the curriculum evaluation models in colleges and universities are still at the stage of looking at them separately,and have not deeply explored the deep connections between them and the potential emotional information in them.Therefore,this dissertation aims to make full use of sentiment analysis technology to deeply mine the rich emotional information of students in the teaching evaluation data,and apply it to the curriculum evaluation model to evaluate teachers’ teaching ability more comprehensively,objectively and fairly,so as to achieve the goal of "evaluation based on evaluation".To promote teaching” and improve the quality of teaching.The main research work of this dissertation has three parts.The first part is based on the characteristics of teaching evaluation text data.This dissertation designs an ATAAE-BERTBi LSTM model based on aspect-level sentiment analysis to obtain students’ different emotional tendencies towards the five teaching evaluation dimensions of the course in the teaching evaluation text.The sentiment obtained by the model The scores correspond to the students’ five-point teaching scoring scores,so the students’ emotional value of different teaching evaluation dimensions can be used to correct the teaching scoring deviation caused by students’ personal scoring preferences.The second part is aimed at the problem that students may have strong personal emotions in teaching evaluation,which may lead to inaccurate and impartial evaluation of courses.This dissertation comprehensively considers the current situation of students’ course evaluation,personal evaluation preferences,emotional tendencies and distribution of course grades.Based on a series of influencing factors,a student abnormal emotion detection model based on XGBoost is designed to identify those students with strong personal emotions.The third part designs the overall scheme of the multi-dimensional curriculum evaluation model.By integrating the support weight matrix,the ATAAE-BERTBi LSTM model based on aspect-level sentiment analysis,and the XGBoost-based student abnormal emotion detection model,the relationship between the scores of seven course evaluation indicators and the five teaching evaluation dimensions of the course is constructed,and provides A visual display of the teacher’s course evaluation results.The data used in this dissertation re real teaching evaluation data.Experiments also verify the effectiveness and availability of the ATAAE-BERT-Bi LSTM model based on aspect-level sentiment analysis and the XGBoost-based student abnormal emotion detection model.The multi-dimensional course evaluation model combines the above two models to correct the teaching scoring deviation generated by students in the course evaluation from the aspect granularity and the student granularity respectively,so that the course evaluation model can more objectively and fairly evaluate the teaching displayed by teachers in the course process.Ability to provide a good basis for subsequent teaching work.