User Emotional Analysis Of Takeaway Platform Data Based On Multi-granularity Analysis

With the continuous development and maturity of takeaway software,an increasing number of people start to order food online.To allow more users to evaluate the quality,taste,and even services of the takeaway products,the takeaway platform has opened a comment function.This function has also gradually become an important reference for users to decide whether to place an order.Therefore,emotional analysis on the comments of takeaway users has become a hot topic widely concerned and researched by scholars.This paper conducted research on real data obtained from the Meituan takeaway platform,used a bidirectional long short-term memory network(LSTM)to extract the time series features in the takeaway comment data.The contextual features were also fully extracted from the takeaway comments with the combination of the self-attention model.On this basis,the multi-model training was conducted on the takeaway comment data to exert the advantages of different network models and achieve better results;finally,based on the above algorithm,a prototype system for emotional analysis of takeaway comment data was developed to provide reference for takeaway merchants,consumers,and operating platforms.The main research contents of this paper are as follows:1.Bidirectional LSTM takeaway comment data emotional classification model based on self-attentionThe bidirectional LSTM network can fully obtain the logical contextual relationship,which is widely used in natural language processing and other fields.To fully obtain the contextual features of takeaway comment data,this paper first built a bidirectional LSTM takeaway comment data emotional classification network.On this basis,the self-attention was introduced to fully explore and learn the relevant laws in the takeaway comment data,making the takeaway emotional classification results more accurate and reliable.2.Emotional classification model for takeaway comment data based on multimodel fusionBased on the data feature extraction by the bidirectional LSTM network,the paper used an end-to-end model with simple operation and fast training speed popular in the field of artificial intelligence for the learning and training of the takeaway comment data.Subsequently,the fast Text model,the BERT-MRC model and other models were used to train the takeaway comment data.In this process,it made full use of the advantages of the fast Text model network such as simple operation and excellent performance,and the advantages of the BERT-MRC model such as better performance,thus strengthening the feature extraction of the takeaway comment data.Consequently,the emotional classification result of the takeaway comment data significantly improved,finally realizing the multi-granularity recognition and classification of the emotional tendency of takeaway comment data.3.The prototype system of automatic emotional analysisTo apply the emotional analysis method for the takeaway platform data proposed in this paper,this paper implemented a prototype system of automatic emotional analysis based on the B/S architecture.The system included functions such as data collection of real takeaway comments,raw data preprocessing,emotional feature extraction and training of takeaway comment data,as well as emotional analysis of takeaway comment data,providing users with a convenient,simple,and easy-to-operate automatic emotional analysis solution.The prototype system were tested through a variety of application examples,proving the effectiveness and practicability of the algorithm and the system in this paper.