Research On Emotion Classification Mining Based On Online Comment Text Of Ele.me

With the continuous development of the current Internet technology and the comprehensive coverage of the Internet in the industry,various commercial platforms and media platforms have accumulated a large amount of text data.The collection and use of this data can help people make better decisions or business development.Different from traditional sentiment classification algorithms,deep learning-based sentiment classification methods can actively learn the semantic information in text from massive data and obtain the text's features and sentiment classification to achieve the purpose of accurately extracting text data and sentiment.At present,research on text sentiment classification has become a key topic in the field of natural language processing.This paper selects the sentiment classification of online review text data of catering online platforms that are rarely involved in the current academic circles.Taking the most representative catering platform in China as an example,it uses deep learning related theories and technologies,literature research methods,and experiments.Analytical methods,comparative analysis methods,etc.are used to build a two-way long-term and short-term memory network model based on the attention mechanism,the ATT-Bi LSTM model,and conduct a comparative study of the models.(1)Through the introduction of previous research literature and deep learning-related theories and technologies,detailed analysis of text data processing technologies,including word segmentation techniques in text preprocessing,part-of-speech tagging,mainstream text representation models,and text feature weight representations Etc.,introduced the word vector technology and introduced Word2 Vec word vector model.(2)In order to make the model pay more attention to the key information in the text,the attention mechanism is explained.At the same time,an LSTM model based on the attention mechanism is constructed to perform the text sentiment classification task,and it is applied to the online review text of Hungry the sentiment classification experiment compares the performance of the six models,including LSTM,Bi-LSTM,ATT-Bi LSTM,SVM,KNN,and Logistic.Experiments show that the ATT-Bi LSTM model makes the accuracy of text sentiment polarity classification tasks further improved.(3)Comparative analysis Based on the effects of LDA topic analysis and cluster analysis on sentiment classification of online review texts.Through the clustering of review texts,we have a clear understanding of the characteristics and keywords of restaurant review reviews.The clustering results adopted a comparatively Visual image,which clearly and intuitively represented the distribution of text clustering.The online review text data of the Hungry Me platform were analyzed based on LDA topic analysis and K-Means clustering analysis respectively,which verified the effectiveness of the algorithm.