Research On Sentiment Tendency Recognition Of Curriculum Reviews Based On Deep Learning

In recent years,with the continuous development of educational informatization,many online platforms have gradually developed,providing a large amount of educational data such as learners’ comments on courses.These review data include star rating data and review text data,which record learners’ overall rating of the course,their difficulties in learning,and their opinions on course resources and instructors.The value of these evaluations is that they can identify the quality of online courses,provide reference for the construction of the platform,and help users choose suitable courses,etc.However,if these data are collected and analyzed by teachers alone,it will take a lot of manpower,and at present,there are star ratings in the review data that are inconsistent with the sentiment tendency of the review text,the scoring data is not highly discriminative,the scores are artificially high,and the reliability is not strong.These problems will affect the overall grade of the course,thus affecting the learners to choose a more high-quality course.Therefore,this paper uses deep learning technology in the field of artificial intelligence to identify the sentiment tendency of a large number of online course review data,evaluate the quality of online courses from the perspective of learners’ perception,and summarize learners’ suggestions for improving the course and the problems encountered in the learning process.Finally,combined with the scoring correction algorithm,we can obtain a course score that is closer to the learner’s true intention.The main work and achievements of this paper include:(1)Summarize the current research status of online course reviews and text sentiment analysis technology at home and abroad through literature review.First of all,it finds the research on online course reviews from the perspective of helping learners to learn independently.Secondly,most researchers pay attention to the text data in the comments and ignore the rating data;finally,there are few studies on the analysis of educational texts using deep learning sentiment analysis techniques in the field of education.Based on the above findings,the research value and innovation points of this study are proposed.(2)Using web crawler technology to obtain the review data of 457 courses in 13 categories such as computer and foreign language in Chinese university MOOCs to form a dataset,and then use deep learning emotion recognition technology to train a sentiment analysis model suitable for educational texts.The models include a sentence-level sentiment analysis model that can obtain learners’ overall emotional attitudes toward comment sentences,and an aspect-level sentiment analysis model that can extract and identify the course evaluation information contained in the comment sentences.(3)On the basis of sentiment analysis,a rating correction algorithm based on sentiment analysis is proposed.The algorithm combines the TF-IDF algorithm to quantify the results of sentence-level sentiment analysis and aspect-level sentiment analysis,so as to revise the course rating.The algorithm can simultaneously correct the incorrectly rated review data with and without aspect words,so that the corrected rating is closer to the learner’s true rating willingness.And the algorithm improves the discrimination and accuracy of the scoring,thereby increasing the reliability of the course scoring and obtaining a more accurate course scoring.(4)Based on the sentiment analysis model and the scoring correction algorithm,and referring to the evaluation indicators in the "MOOC Quality Evaluation Form" issued by the UOOC Alliance,a course feedback model is constructed.And obtain the review data of 4 courses of MOOC of Chinese universities outside the data set as the verification data to verify the application value of the course feedback model.Based on the analysis results of each course,a targeted improvement strategy is proposed for a single MOOC course,so as to provide a reference for the construction and reform of various MOOC courses and course selection for learners.