Design And Implementation Of Unattended Supermarket Recommendation System Based On User Features

"New Retail" is the application of new thinking and new technologies in the Internet field to traditional commodity sales.Therefore,as a representative of "new retail",unmanned supermarkets must supported by advanced technology of Internet to enable consumers to enjoy a more comfortable and efficient shopping experience.In order to implement the system,the main research works are summarized as follows:(1)Aiming at the problem that the unmanned supermarket recommendation system can't obtain user features through user registration information like the e-commerce platform,this paper uses face feature recognition technology,and finally realizes real-time acquisition of user features by the recommendation system through image preprocessing,feature extraction,classification training and other links.Among them,the designed gender recognition method has improved the accuracy by more than 0.3% compared with the traditional method.(2)This paper designs and implements a recommendation system that meets the needs of unmanned supermarkets.First,Aimed for characteristics that users' preference for goods of the same characteristics are similar.In this paper,user characteristic factors are combined into the model of commodity recommendation algorithm based on user collaborative filtering algorithm.Secondly,aiming at the problem of high requirement of timeliness of the recommended goods for the users of unmanned supermarkets,this paper divided and optimized the shopping record of the users in time periods,and constructed a more accurate user-commodity evaluation matrix.Thirdly,in order to meet the demand of unmanned supermarkets to improve sales volume by Means of promotion,the k-means algorithm is used in this paper to design the commodity promotion module of various promotion methods.Fourthly,based on the model of commodity recommendation algorithm and the model of commodity promotion,this paper designs the overall and functional modules of the system.Among them,it uses sequence diagram to describe the information transfer process between modules.Finally,this paper uses the service of ZooKeeper and Dubbo framework to complete the implementation of the unmanned supermarket recommendation system and pass the test.(3)In the application system of the unmanned supermarket recommendation system,a series of devices are needed to realize the specific work.Therefore,this paper designs two devices for the recommendation system.Firstly,this paper uses Hi3518 E.The chip uses multi-threading technology to design a feature recognition device that can realize real-time acquisition,recognition and transmission functions.Secondly,this paper designs a mechanism based on the recommendation system of the vending device,and selectts the key components through calculation.The test results show that the unmanned supermarket recommendation system designed in this paper not only improves the performance of the algorithm compared with the traditional algorithm,but also has a good application effect.Therefore,the system can play a role in improving the sales of unmanned supermarkets and increasing the user experience.