Design And Implementation Of Commodity Review Analysis System Based On Sentiment Analysis

With the development of Internet technology,online shopping has become the first choice of most consumers,and more and more merchants have begun to open up online sales channels,which is both an opportunity and a challenge.In order to stand out from the highly competitive sales environment,obtaining and analyzing the feedback information in the product review data is a very effective means.Merchants can use the feedback information to improve and promote the product to increase sales.This paper implements a set of speed insight system based on sentiment analysis.The system can configure the collection items through the visual interface,and then use the data collection module to collect data from the e-commerce platform to build a product review corpus.At the same time,two sentiment analysis models are also embedded to perform data analysis on the processed review data.The layered sentiment model is responsible for performing coarse-grained sentiment analysis on the review data to obtain product reputation,and the review theme model is responsible for performing fine-grain sentiment analysis on the review data to obtain product review topics.The background and research status of the design of the super speed insight system are first introduced.Then introduced the related theory and key technology involved in the system.Next,the functional and non-functional requirements of the system are analyzed,and on this basis,a summary design is performed to define the system architecture,module functions,and database design.Afterwards,the three major modules of the system's configuration management interface,data collection,and data analysis were designed and implemented in detail.Two emotion analysis models required by the system were constructed.The two-layer emotional tendency model uses CNN,bidirectional LSTM,and attention Build and comment topic models are built using LDA models.Finally,perform functional testing and performance testing,and summarize and prospect the work done at the end of this article.