Recommendation System for Multivariant E-Commerce Interfaces

The main objective of e-commerce activities is to sell products or services and possibly provide after-sales service. Depending on how this is achieved, various methods of personalizing communications are employed. Customized content or design is sometimes practiced in online shops, B2B sales platforms, auction portals, or marketplaces. Some popular approaches to personalization in e-commerce include product recommendations, dedicated ads, or emails. Occasionally, other solutions are also implemented based on business owner decisions. Most current solutions primarily focus on personalizing content, and the ability to provide a holistic dedicated user interface is less commonly utilized. However, the potential of personalizing layouts by offering multiple UI variants in e-commerce seems underestimated. Undoubtedly, such an approach requires a system architecture design capable of meeting potential users’ requirements. The proposed solution should be modular and scalable, minimizing barriers to entry. A possible starting point for developing a framework to serve dedicated user interfaces could be the recommendation systems used in today’s e-commerce. This chapter discusses the applications of recommender systems and introduces various approaches to using multivariant user interfaces in e-commerce. An analysis of typical modules of recommender systems is provided, with a particular focus on components critical for delivering dedicated layouts. A solution architecture is proposed that enables the practical implementation of the concept of serving user interface variants tailored to the decisions and behaviors of specific groups of e-commerce customers.