Design And Implementation Of Commodity Evaluation System Based On User Comments

With the rapid development of Internet shopping, the rapid rise of the logistics industry, the continuous improvement of online payment function, online shopping has become the main way of shopping. Online shopping is not only convenient, but also to provide users with commodity comments function. People can comment on the whole or some features of the commodity through comments, as the opinion and feedback. And commodity comments can provide consumers with more abundant commodity information. And reading commodity comments becomes a very effective way to understand the details of commodity. At Present, the information of commodity provided by the e-commerce websites is not comprehensive enough. And each commodity has a large number of comments. People need to spend a lot of time to read a large number of comments in order to fully understand the information of the commodity and information of the features they are concerned about.In order to solve the problems caused by the large number of commodity comments, this paper designed and implemented the commodity evaluation system that can provide users with commodity evaluation information, including the sentiment classification of comments, features of the commodity and the sentiment classification of features.The main work of this paper is as follows: Firstly, this paper studies and briefly introduces the technologies and algorithms used to implement the commodity evaluation system, including web crawler technology, word vector and three algorithms for sentiment classification: method based on dictionary and syntax analysis, Naive Bayesian, convolutional neural network. Secondly, this paper makes a detailed analysis of the requirements, and introduces the hierarchical structure of the system and the division of functional modules. At the same time, this paper designs the database based on the functional requirements, and the overall interface design of the system is introduced. Then, this paper introduces the design and implementation of the core function modules of the system. The core function modules mainly include:the feature extraction, sentiment classification and commodity evaluation. Finally, the performance of the three kinds of sentiment classification models is evaluated, and the accuracy of the evaluation results show that the evaluation information of the system provided by this paper is accurate.