I. Introduction
Recent years have witnessed continued efforts in improving the effectiveness and efficiency of recommender systems. Such systems have been an essential component in helping customer to discover their desired items and assisting business owners to earn more profits, ranging from movies, news, to point-of-interests (POIs). Unlike these applications, user consumption behaviors in E-commerce are even more complex, which are usually influenced by many product’s characteristics. This makes it difficult to accurately model a user’s true preference. Naturally, a question turns out: what affects the user’s purchase decision making? Namely, which aspects of a product drive the user to purchase the product?