A location-based orientation-aware recommender system using IoT smart devices and Social Networks

Abstract

The rapid development of IoT sensors and data provided by Social Networks has necessitated the fast development of recommender systems as they can be used as a tool to filter items that are more likely to be preferred by users. A major goal of recommender systems is to provide users with personalized recommendations after analyzing their preferences. IoT smart devices and Social Networks have opened windows of opportunities for user preferences to be dynamically recognized. Although analyzing user preferences helps to provide more personalized recommendations, considering location and orientation information as user contextual information results in more relevant recommendations being provided. Location information has been used by Social Networks, especially Location-Based Social Networks, in order to provide recommendations based on current user location. However, the importance of the user orientation context has been overlooked by almost all of the research done in this area. Developing a location-based orientation-aware recommender system can perfectly bridge this gap. For this study, a location-based orientation-aware recommender system is proposed as an innovative type of recommender system. The proposed recommender system is able to not only apply contemporary user contextual information to the recommender algorithm, but also makes progress towards preparing more personalized recommendations by taking user orientation context into account. For this study, user preferences are dynamically measured by IoT smart devices such as smartphones, Google Home, and smartwatches. Information provided by virtual communities extracted from Social Networks helps the recommender system in situations in which user preferences are not extracted from their IoT devices. In addition to user preferences, their smartphone pointing direction has also been applied as their orientation context for the recommender algorithm in outdoor environments. To evaluate the impact of the user pointing direction in our proposed methodology, an event recommender system based on the real data was implemented and examined in the city of Tehran in Iran. Because of the challenging nature of social events, a simulated experiment is also presented for the City of Calgary. Also, the system results are compared with the results of Collaborative Filtering and Content-based recommender algorithms to demonstrate the power of the recommendation engine. The evaluation indexes prove that our proposed recommender system outperforms its counterparts by providing more accurate and personalized recommendations.