Master's thesis

Annotation

With the high growth of online shopping platforms and the increasing number of products on offer, the need to provide customers with relevant and tailored product suggestions has become crucial not only to improve the user experience but also to increase sales for many e-commerce companies. In today's competitive environment, where customers expect a personalized approach and quick finding of desired items, it is essential that e-shops use advanced methods and technologies such as artificial intelligence. These technologies enable the analysis of customers' purchasing behavior, on the basis of which they can identify and predict needs, which leads to the creation of personalized recommendations. In our work, we designed and implemented an approach based on sequential data processing, which combined predictions from models reflecting the purchasing behavior of specific users with a model reflecting the interactions of all users. This hybrid approach allowed a better understanding of the individual preferences of a particular user while leveraging broader behavioral patterns across all users. This approach resulted in more accurate recommendations generation compared to a model reflecting the purchasing behavior of all users used alone.