Research On Topic Mining Of Health Air-conditioners Online Reviews Text Based On Supernetwork Model

Online reviews,as product usage feedback information,not only help consumers understand the quality of product features,but also provide service and product improvement directions for merchants and manufacturers.In the post-COVID-19 era,the demand for health air-conditioners has grown against the trend,and users’ awareness of healthy consumption has risen.In order to improve product user satisfaction,it is very necessary to deeply mine valuable information such as users’ subjective emotion and focus on product features and services.However,in the face of the rapidly growing and diverse online reviews of health air-conditioners,relying solely on manual analysis and processing is inefficient and cannot resolve the contradiction between text processing and rapid decision-making in the era of big data.How to realize automatic and intelligent text processing through effective mining methods has become a hot research topic in this field.This paper takes "health air-conditioners" as the research object,and follows the main line of "online reviews text sentiment classification-binary classification text topic mining",proposing a health air-conditioners online reviews text topic information mining model.First,combined with the characteristics of health air-conditioners online reviews text,the collected user reviews are subjected to text preprocessing,equalization processing and vectorized feature extraction.Establish an sentiment classification model Tc Bi LSTM-SA based on two channel bi-directional long short-term memory network(Bi LSTM)and self-attention mechanism.By comparing the sentiment classification results of the Tc Bi LSTM-SA model and other models,the optimal model performance evaluation index values are obtained in different model comparison experiments,two channel feature adjustment experiments and self-attention adjustment experiments,which verifies the advantages and effectiveness of the Tc Bi LSTM-SA model.Then,a four-layer review network based on supernetwork and topic oriented-health air-conditioners product feature topic supernetwork model is constructed,and the high-frequency feature words and opinion words in the reviews text and titles text are extracted by using the topic model(LDA).Design four methods of health air-conditioners reviews text topic information mining,and feasibility analysis of the topic mining model which to make the mining results more convincing.Finally,select the Haier-3P and TCL-big 3P health air-conditioners in Jingdong,and use the Tc Bi LSTM-SA model to classify the sentiments of the two health air-conditioners online reviews into positive reviews and negative reviews.Through information mining,we obtained the main advantages and selling points of two types health air-conditioners,the hotspot features that users focus on when purchasing air-conditioners,the comprehensive product feature performance evaluation table,and the credibility of the title product features,which verified the validity and feasibility of the constructed sentiment classification model and topic mining model.Analyzing the mining results,making rationalization suggestions for consumers.This thesis has 42 figures,25 tables and 100 references...