Gift finder : a Web 2.0 recommendation system using mobile technologies

Abstract

One of the challenges facing recommendation systems today is the need to base recommendations on a User profile, which in most cases is assumed to be about the user using the system. What happens however when we require recommendations for someone else? What happens when our own profile is used to generate recommendations for a product that is not meant for us? GiftFinder attempts to answer these questions by presenting a different approach to product recommendations; one which relies on the information available on Social Networks. Through such networks we attempt to make use of evolving data that is not just dependent on previous purchases but on the entire social experience online. Moreover, through the social graph of friends, we are able to build interest profiles for anyone in the social graph, rather than being limited to our own profile. GiftFinder makes use of a number of Web 2.0 technologies to experiment with this idea, such as Wikis and Concept knowledge bases. Moreover, the prototype itself embraces the world of Mobile technologies, with the system developed to be used on a mobile device in conjunction with a shop server and product database so as to make the experience a portable one. Our interface is therefore light and simple so that by simply choosing a friend, the user can obtain recommendations through a Bluetooth connection with the Server in the hope of aiding the purchasing process. We have conducted an evaluation with this prototype using a fairly small pool of evaluators. While initial conclusions cannot be drawn because of this small sample size, we have shown in our discussion that the approach, while far from mature, is promising and meets the aims set out at inception. This is backed with a majority of positive user feedback and therefore we hope that this system contributes to the area and provides a step pins stone for further development of such system