E-Commerce Recommendation System Using Mahout
Phan Duy Hung, Dinh Le Huynh · 2019 IEEE 4th International Conference on Computer and Communication Systems (ICCCS) · 2019
The last two decades has witnessed the speedy growth of e-commerce, most conspicuously in online shopping websites such as Amazon, eBay or Alibaba. The popularity of e-commerce increases the users' lifetime value and makes online market more competitive. The strategy for platform owners is to make highly customized suggestions to customers meaning that they should know users individually by learning from their habits. Different from a traditional brick-and-mortar store where there is only one store for all users, in e-commerce, if you have 3 million customers on the Web, you should have 3 million stores on the Web (Jeff Bezos, CEO of Amazon). Therefore a need for deeply understanding users to create personalized recommendations is formed.The basic idea of recommendation is linking one product (item) to other products, one user (client) to users, as clusters, and then create connections between them. Initially, it meant just suggesting the same purchased item to other buyers. The state of affairs has been gradually improved by the technology and effort of researchers. Nowadays, Web recommendation system has become an essential part of all online e-commerce systems. In this paper, a novel process for constructing a recommendation system is proposed. And we explore the recommendation algorithm through a case study of Web log mining a real-world online shopping site.