Session-based Personalized Recommender System for Online Shopping

The recommendation system (RS) plays an important role in aiding e-commerce customers in overcoming information overload. In the last decades, the recommendation system has made significant advances. In specific, user-dynamic behavioural context concerning the item. Modelling user-behavioural context is complex and has been employed by state-of-the-art methods. However, this model fails to address the several problems occurring in future real-time online shopping portal environment. This paper aimed at addressing the issues of online recommendation considering continuous data (i.e. streaming environment) in an e-commerce environment. Customers and products may rely on a certain unexpected scenario and can arise variously at varied rates. This paper presents an efficient RS by continuously identifying (i.e. using session-specific information) and adapts to changing scenarios in customer items views, customer favourites (i.e. preferences) and product portrayals. Further shows the significance of proposed time-centric prediction/recommendation (TCP) when compared with the existing recommendation model.