Design And Implementation Of A Customer Feedback System Based On Maximum Entropy Model

In recent years, with the development of e-commerce, the quantity of customers’ online comments explosive increase. Many researches for the affect of customers’ online comments on the performance of products were studied, has been proved that customers’ online comments significant impact on the sales of their products and services. Many business pay more and more attention to customers’ online comments now,and to do something for the impact of negative comments positively. Wihle only part of the problem could be solved, because the enterprises have not get customer satisfaction of current products and new requirement by analysing those comments. It is very difficult to handling a great deal of customer’ comments by manual or simple tools. Whereas a solution for analysing customers’ comments is proposed in this thesis.In this thesis, the core demand points are that mining customers’ satisfaction with current product and new requirement. Firstly, obtain from the customer e-mail, visiting records and other ways, as the original data systems; Secondly, pretreatment of raw data, including text segmentation, filtering and other operations; Then classify those data according to company’s product; calculate the tendencies of each feedback information by emotional dictionaries and maximum entropy algorithm; and finally the system summarize the results of the analysis presented to the user, provide reference data for business development in future.The topic is a actual project developed for internal staffs, the author participated in analysing requirement and the design and implementation of main function of the project. After analysing the requirement of the project, we described the solution of judging emotional tendencies in more detail. In the end of the thesis, the author describes the system architecture design and the process of main module design and implementation in detail.During the testing phase and the actual use show that the system could complete the task of feedbacks classification and the tendency judgment, and show users an intuitive reference data. Internal staffs improve work efficiency with the system, which show that we achieve the desired purpose.