A Course Recommendation Algorithm Based On Sentiment Analysis Of Pop-up Text Comments

Online course recommendation tasks have strong application value in several aspects such as learning path planning,course quality assessment and global education resource sharing.Recommendation algorithms based on sentiment analysis have achieved remarkable results in the recommendation field in recent years.The development of online education has given rise to the emergence of pop-up text reviews with characteristics such as real-time,cross-discipline and diversity.In response to the problem that comments in course pop-ups that are not related to the course scene do not contribute much to the course recommendation task and that traditional LSTM and CNN models cannot extract both bidirectional long-range features and local features of the text,this paper conducts an in-depth study.Firstly,a sentiment analysis model based on a fused conditional random field attention mechanism is proposed in the sentiment analysis task,and secondly,a deep learning recommendation model based on Bi GRU-CNN-Attention-FM is proposed in the recommendation task.The main work and contributions of this paper are as follows.First,A sentiment analysis model based on a fused conditional random field attention mechanism is proposed.Firstly,the ALBERT model is used to perform dynamic word embedding on the input sentences to obtain the word vector of each word in the context and enhance the model’s understanding of the text.To address the problem that comments in course pop-ups that are not related to the course scenario do not contribute much to the course recommendation task,an attention mechanism incorporating conditional random fields is proposed,enabling the model to filter out as many sentences as possible that match the target topic.Finally,the feature vector characterizing the sentiment polarity of the sentences is connected with the feature vector characterizing the topic of the sentences,and the final sentiment polarity is output.Comparative experiments show that this method improves 1.4% in accuracy and 2.6% in F1 value index over the BERT-ATT-Bi LSTM-CNN model for the crawled pop-up dataset.Second,a recommendation model based on Bi GRU-CNN-Attention-FM(BCAF)is proposed to address the problem that the pop-up text is rich in contextual information but traditional models cannot extract both local and bidirectional global features of the text,this paper introduces the Bi GRU model,firstly,the input layer receives the time series features and keyword features outputted by the previous model and filtered manually,and then captures the forward and backward information in the sequence through the Bi GRU model.The Bi GRU model is then used to capture the forward and backward information in the sequence;due to the different importance of the information contained in the words in the input sequence,the attention mechanism,which is more common in deep learning models,is introduced so that the model can focus on the key areas of the text comment and better capture the key information in the text comment;finally,a convolutional neural network model is introduced to make the Finally,a convolutional neural network model is introduced,which enables the model to have its local feature extraction capability.Comparative experiments show that the accuracy of this paper’s method in the crawled pop-up dataset is 1% better than that of the Deep Co NN model,and the mean square error is reduced by 0.0145.Second,a recommendation model based on Bi GRU-CNN-Attention-FM(BCAF)is proposed to address the problem that the pop-up text is rich in contextual information but traditional models cannot extract both local and bidirectional global features of the text,this paper introduces the Bi GRU model,firstly,the input layer receives the time series features and keyword features outputted by the previous model and filtered manually,and then captures the forward and backward information in the sequence through the Bi GRU model.The Bi GRU model is then used to capture the forward and backward information in the sequence;due to the different importance of the information contained in the words in the input sequence,the attention mechanism,which is more common in deep learning models,is introduced so that the model can focus on the key areas of the text comment and better capture the key information in the text comment;finally,a convolutional neural network model is introduced to make the Finally,a convolutional neural network model is introduced,which enables the model to have its local feature extraction capability.Comparative experiments show that the accuracy of this paper’s method in the crawled pop-up dataset is 1% better than that of the Deep Co NN model,and the mean square error is reduced by 0.0145.The sentiment analysis and recommendation model proposed in this paper makes full use of the pop-up text information posted by users during the online course learning process to better solve the course resource recommendation problem...