A store location-based recommender system using user’s position and web searches

ABSTRACT Nowadays, gathering information about commercial products can be performed either online or offline. While online searches can be virtually undertaken through online shopping websites, offline searches should be done physically at stores. However, there is a specific emerging trend where users can check some product opportunities online before getting to the stores and then possibly buying some items whose properties have already been evaluated over the Web. Product properties can be studied online while on site evaluation provides a direct contact with these goods at the stores. The objective of the approach developed in this paper is to discover user preferences when searching and exploring online shopping websites and then recommend the stores that better match their interests. First, users’ internet behaviours are extracted from an online shopping website. Secondly, a Voronoi high-dimensional structure supports the derivation of similarities between the users and stores. Third, a distance matrix between the user and the selected stores is generated. Finally, a ranked list of the most appropriate stores is provided to the users based on their product interest and their locations. The whole approach has been successfully tested by a panel of 30 volunteers in the 6th District of the city of Tehran.