Design And Implementation Of E-commerce Commodity Session-based Recommendation System With Heter-Ogeneous Graph Neural Network

In the era of big data,massive data information has caused a serious problem of "information overload" for users.Recommendation system is one of the methods to overcome this problem.In the field of e-commerce,the recommendation system can captures users’ preferences by mining the historical interaction between users and goods,and recommend goods that users are interested in,creating greater profits and value for e-commerce platforms.Compared with traditional recommendation methods,which only focus on users’ long-term preferences,session-based recommendation methods decompose user behaviors into sessions and consider the transaction structure of user behaviors,which can timely capture the transfer of users’ interest preferences and show more reliable recommendation results for users.However,previous studies on sessionbased recommendation often focus on the sequential relationship and transformation relationship between items in the session-based dataset,ignoring other important information.Therefore,a session-based recommendation algorithm with Heterogeneous Graph Neural Network is proposed in this paper.The algorithm constructs the session dataset into a heterogeneous graph containing three types of nodes:user,good and session.It uses Heterogeneous Graph Neural Network to capture the complex dependencies between nodes,and emphasizes the main purpose of the current session combined with Attention Mechanism.In order to verify the performance of the model,this paper compares it with the popular recommendation methods on two open-source datasets.At the same time,this paper designs and implements an e-commerce commodity recommendation system with the separation of front and back ends.Different from the traditional recommendation system,this system takes the session as the basic data unit,and applies the algorithm of sessionbased recommendation with Heterogeneous Graph Neural Network to the recommendation engine of the system,which provides the current sessionbased recommendation service for users.The front end of the system is mobile APP,which is developed with Android.And the back end uses the Flask framework to build Web services,completing the development of user management,good management,session management,good recommendation and visual analysis function modules.