基於社會網絡關係之推薦系統-以網路寵物用品業者為例

Abstract

網際網路蓬勃發展加上許多社會網絡或社群平台逐漸興起，不僅串接人與人間的社交行為以及相關社會資訊分享，就連傳統電子商務的商業交易行為，也慢慢移動到此社會網路或社群平台，形成了社群商務或社會網路商務。社群商務與傳統電子商務最大不同點，在於消費者購買行為很容易受到社群朋友的影響，因此應結合社群網路的朋友關係資訊，加上傳統電子商務所蒐集到的交易行為資訊，來形成一個有效的消費者社群推薦模式。本研究提出一個結合進銷存系統中，記錄的顧客歷史購買產品資料及Facebook社群或社會網路平台中的朋友關係資料之社群推薦架構與流程，並透過網路寵物用品經營業者的資料，實證本研究提出社群推薦架構與流程的可行性與有效性；此外，亦發現考慮平均利潤推薦策略，不會增加成功推薦商品的比率與推薦精確度。

In recent years, the increasing popularity of social networking technologies and platforms has opened up a new era of electronic commerce, called social commerce, which changes the traditional thinking about online shopping. Social commerce uses the information such as user rating, social advertising and social friend relationship of social networks to assist in the buying of selling of products. The major difference between electronic commerce and social commerce is that consumers often rely on the advice and recommendations from online friends when making purchase decisions. Therefore, the traditional recommendation mechanism in electronic commerce is not suitable for the social commerce. This study proposes a social recommendation system that can generate personalized product recommendations based on the transactional purchase records from the legacy systems and social relationship from the social networking platforms, Facebook. Accordingly, our experiments employ the case study of an online pet shopping to show that the proposed social recommendation system outperforms traditional recommendation mechanism in terms of the precision and recall. The proposed social recommendation system can also be effectively applied to social-commerce retailers to promote their products and services in the social networks.