Content area
Full Text
Abstract
Recently, a vending machine sells various items and it is convenient to manage because of automation system. In addition, it is equipped with a display so it is useful for changing merchandise data. However, due to the display limitation, sometimes none of the items can displayed on a screen. In addition, it shows the same home screen to all users, users can feel inconvenience if they cannot find the item which they want on the screen. Therefore, it is necessary to find a way that users can find what they want quickly by using personalized recommendation service. Smart vending machines, which is combined with IoT technology, can distinguish users through their smartphones. In addition, it provides recommendation considering the preferences of users that change with different circumstances.
Key Words: Smart Vending Machine, Recommender system, Personalized system, Context-aware system, IoT(Internet of Things)
(ProQuest: ... denotes formulae omitted.)
1.Introduction
Recently, as the number of users of smartphones and social network services (SNS) have increased, a personalized recommendation system has emerged that can provide highly accurate recommendations. In particular, many researches carried out to provide a convenient service in everyday life by combining with IoT (Internet of Things) technology, and various types of IoT devices have released to improve the quality of our lives [1].
The smart vending machine is a system formed by combining IoT technology and vending machine. The vending machines are performing monitoring functions such as determination of stock replenishment criteria, equipment failure, etc. with focusing on convenience of management [2]. However, there were difficulties in predicting sales volume and inventory because users' choices could be different depending on the season or the weather. In addition, existing vending machines have a fixed home screen, which is inconvenient for users to find what they want. In such problems, a more convenient service can provided to the user if it could predict what user wants in advance and recommend the product information.
Most of existing recommender system uses the purchase history of the user [3]. In addition, the existing recommender system finds users who have similar histories or preferences, and improves the accuracy of recommendation through the user's history [4]. However, there is problem that always shows the same recommendation list because the user's preference...