Review Analysis Technology And Application For MOOC

With the rapid development of Internet technology and the increasing demand for high-quality talents in the society,MOOC,as one of the most important application innovation products of "Internet + Education",plays an important role in the field of online education.But there are also serious problems in the development of MOOC.The main problem addressed by this article is that the current teacher-student ratio is very different.Without big data analysis methods,it is difficult for teachers to respond to student feedback in a timely manner.In response to the above problem,this thesis conducted the following research on how to automatically analyze MOOC reviews:First,this thesis designed an extraction model based on multi-task learning.The model can simultaneously extract the aspects and opinion words in the comments,and classify the sentiment polarity of the aspects.Through multi-task learning,the model avoided the error cascading caused by the pipeline approach,and used the connection between the various tasks to improve the effectiveness of each task.Through comparative experiments,the model has achieved the best results on multiple standard English datasets.Secondly,because the extraction model mentioned above was trained through supervised learning,the quantity and quality of the labeled data are relatively important.In fact,the labeled data of the MOOC reviews is seriously insufficient.Therefore,this thesis proposed a network-based migration algorithm and an adversarial-based migration algorithm,which aimed to learn general knowledge from a large number of labeled product reviews,in order to improve the effectiveness of MOOC review analysis in the case of insufficient data.The models of both migration algorithms were based on the multi-task extraction model proposed above.This thesis analyzed the migration effects of these two algorithms through comparative experiments in order to choose a better migration algorithm in practical applications.Finally,this thesis applied the results of the above algorithm research to the design of mobile educational assistant system.The system was designed to help MOOC teachers analyze and summarize student feedback quickly and easily.Its specific function was to provide objective comprehensive evaluation text for a given MOOC review data.In order to classify the objects obtained by the extraction model,this thesis proposed a classification method based on vector similarity;this thesis used text replication to generate comprehensive evaluation text.Through the above work,this thesis basically solved the problem that MOOC teachers are difficult to timely get the feedback in the student reviews.