Po-Yen WuYu-Rong LaiRAY-I CHANGShin-Yat Liu2024-10-182024-10-182019-11-189781728141176https://www.scopus.com/record/display.uri?eid=2-s2.0-85205281136&origin=resultslisthttps://scholars.lib.ntu.edu.tw/handle/123456789/722224Customers regard good user-experience (UX) as the key factor in online shopping. Therefore, e-commerce (EC) websites will track users' behaviors for analyzing their relations to products. According to these analysis results, the websites' layout/workflow and recommendation are tuned to improve UX. However, conventional analysis tools usually focus on user-to-product relations. They are not good at analyzing "big and graphical"user behavior data. In this paper, we develop a graph analysis system with deep learning GNN (graph neural network). Our system uses the standard format xAPI to record user behavior data. We use graph analysis to discover their relations for improving UX. Experiments are on a real EC website with about 6,000 users in 3 months. Our system can reduce over 28% browsing actions of users to reach the page of desired products.falsebig datadeep learninge-commercegraph analysisgraph neural networkUXxAPIconference paper10.1109/DataCom.2019.000422-s2.0-85205281136