Design And Implementation Of Retrieval System For Product User Reviews

With the increasing development of the Internet and big data,online shopping plays an increasingly important role in people's daily lives.Information retrieval has become an effective method for online shopping for product selection.In recent years,online shopping surveys on consumers have found that consumers pay most attention to such information as user reviews of goods when shopping.User reviews of goods can specifically reflect consumers' actual feelings about the goods and help to judge the overall quality of-the goods.This thesis aims to build a retrieval system for product user reviews.The system can effectively use user reviews of commodities,retrieve product reviews through text or images,and assist users in better purchasing commodities.Specifically,in the face of actual application scenarios,this paper combines Web application development technology and machine learning technology to design and implement a complete,convenient and practical product user comment retrieval system.The system comprises six main functions:registration/login function,user information management function,comment retrieval function,data collection function,data analysis processing function and system management function.This thesis retrieval system consists of two subsystems:Web subsystem,data collection and processing subsystem.The Web subsystem is responsible for the interface with users,providing users with functions such as registration,login,information management,and interface display.It comprises three modules:website front end,website back end and data storage.The Web subsystem uses the processing data of the data collection and processing subsystem to perform its work.The data collection and processing subsystem is responsible for acquiring product data from the network to perform text or image-based review retrieval.It includes two modules of data collection and data processing.The data collection module crawls the product pictures and corresponding comments from the e-commerce platform,and then passes it to the data processing module.De-redundant processing of product reviews,and then use a recurrent neural network to build a sentiment analysis model to determine the sentiment polarity of reviews.A binary tree algorithm is used to establish a comment retrieval method for searching images by text;a convolutional neural network is used for feature extraction;a K-Means clustering algorithm is used to establish a comment retrieval method for searching images by image.The data collection and processing subsystem provides application data for the Web subsystem.Tests show that the system can complete system functions such as user registration/login,user information management,and system management,as well as emotional sentiment analysis of product reviews.On this basis,it can respond to retrieval of user reviews based on images or text.