Location Aware Product Recommendations

Abstract

Nowadays a typical retail chain store catalog encompasses thousands of products, the sheer quantity of products makes it dicult for the customer to be familiar with all the options and their specificities without spending too much time in each shopping trip. In order to make products known that the customer may be interested, while providing potential store sales, recommendation systems are applied to reduce the information examined by the customer and help him decide alternatives, to explore other products and categories that may please him. With the vast customer knowledge that stores already keep, it is possible to extract information such as preferential products, shopping patterns, product related categories and thus understand what can make a better shopping experience for the customer. Recommendation systems can be applied to any store type, usually traditional recommendation systems based on collaborative or content-based filtering use simple models. Context-aware recommenders take into account not only the customer purchase history but the context in which those purchases were made, and also takes into account the target user current context when generating recommendations. One possible context is the user's location and whereabouts inside the store, with this type of information it is possible and desirable to use it to produce better, more personalized and timely (well-timed) product recommendations. The final product of a recommendation system should be considered as a powerfull personalized assistant who knows the customers and all the products of the store, and during their shopping trips, advises and guides them according to their tastes and in this case their location. Taking advantage of Fraunhofer AICOS previous experience and know-how in the areas of accurate internal location and product recommendation, these two techniques were combined into an innovative solution that helps improve customers planning and shopping trips offering counselling before and during the customer journey. Context-aware recommendation systems was explored combined with periodic and sequential pattern mining in order to build a robust shopping companion app and support system.