Research On Aspect-based Sentiment Analysis For MOOCs Reviews

With the development of Internet technology and the progress of society,more and more learners improve their abilities through online learning,and a large number of MOOC course reviews have been generated.Aspect level sentiment analysis method can analyze aspect words and sentiment tendencies in review text,but at present,this method is mainly used in the field of e-commerce and hotel catering,and the application of aspect level sentiment analysis in the field of MOOC course review is relatively rare.Therefore,this paper aims to analyze the aspect words and sentiment tendencies in MOOC course reviews by combining MOOC course evaluation indicators and sentiment word information.Rule-based and deep learning methods were used to analyze aspect words and sentiment tendencies in MOOC course reviews.The main work of this paper is as follows:(1)Use a rules-based approach to analyze aspects and sentimental tendencies in MOOC reviews.This paper first studies the syntactic relationship between sentiment words and aspect words,and summarizes the rules of aspect words combined with the syntactic dependency analysis tool.Then,the irrelevant aspect words are filtered according to the word vector of the course evaluation information.Finally,the results show that the rule-based method can make full use of emotional word information to identify subjective words in the reviews,but the utilization of course evaluation information is low,and the recognition effect of objective words is relatively poor.(2)Use deep learning method to analyze aspects and sentiment tendencies of MOOC course reviews.Based on the BERT model,this paper adds the sentiment word focus layer and the MOOC course evaluation index layer.In this paper,additional sentiment word information is introduced,the weight of the local context of sentiment words is dynamically adjusted according to the co-occurrence relationship between sentiment words and aspect words,and the model focuses on the local context of emotional words by combining the attention mechanism to improve the accuracy of the model.At the evaluation index layer,the MOOC course evaluation index is set as a query vector,and the attention mechanism is combined to make the model focus more on the aspect word features related to the task.The results show that the improved BERT model can effectively use sentiment word information and MOOC course evaluation index information,and has a certain improvement in terms of aspect word recognition and emotional polarity classification tasks.This paper combines sentiment word information and MOOC course evaluation information,and uses syntactic dependency-based methods and deep learning methods to identify aspect words and sentiment polarity in MOOC course reviews,respectively.The results show that sentiment word information and course evaluation information are helpful in identifying fine-grained aspects of and sentiment polarity in reviews.