推薦系統已成為大多數電子商務網站中的強制性組件。除了傳統的推薦系統之外,許多平台(例如Amazon,Steam或Alibaba)也開始在其推薦模型中應用捆綁策略。捆綁技術是很久以前在營銷領域研究的熱門話題。創建捆綁包最實用的技術之一就是將高度相關的產品組合在一起。無論如何,從該領域開發的大多數技術通常被認為是監督技術。這導致需要領域專家。在這項研究中,我們提出了深度圖卷積神經網絡體系結構,以根據用戶的即將到來的目的地和最近的會話自動生成個性化的度假體驗產品包。我們將此問題視為多標籤問題。然後,我們將模型應用於來自KKday的數據集,KKday是一個提供度假體驗產品的平台。
The recommendation system becomes a mandatory component in most of the e-commerce websites. Apart from the traditional recommendation systems, many platforms such as Amazon, Steam, or Alibaba start to apply the bundling strategies in their recommendation model. Bundling technique is a well-studied topic from the marketing field since a long time ago. One of the most practical technique to create a bundle is to group highly related products together. Anyway, most of the techniques developed from this field usually considered as supervised techniques. This resulting in the need of a domain expert. In this research, we proposed the deep graph convolution neural network architecture to automatically generate a personalized bundle of holiday experience products to the user according to their upcoming destination and their recent session. We treat this problem as a multi-labeling problem. We then apply the model to the data set from KKday which is a platform provided a product for a holiday experience.