Certain investigations on performance analysis of personalized recommendation using graph neural network

Abstract

In the digital era, people want to look at the world through the lens of interests, behaviours, preferences, and experiences. Since the internet has become a hub of product, information about the product is overwhelming which makes the customers displeased. In this age of information overload, adding an element of personalization to the product can go a long way in engaging the customers in a pleased way. This scenario can be easily handled by a recommendation engine which helps you to deliver targeted, personalized solutions to the users. Moreover, the recommendation system is majorly classified into collaborative filtering, content-based filtering and hybrid filtering. These three types of recommendation models use techniques like text mining, KNN, clustering, association rule mining, matrix factorization, neural networks etc., to predict similar items to the users. Now a-dayand#8223;s research is being done on deep learning algorithm to improve the effectiveness of the recommendation system. Also, these deep learning algorithms are used along with the graph data structure to improve the expressiveness and fidelity of system. The proposed research addresses the problem of providing personalized recommendations to the users of the system using graph based deep learning techniques. Particularly, there are three major challenges in the recommendation system. The first challenge is how to handle heterogeneity and sparsity. The second major challenge is incorporating the graph structure information along with the context information in the recommendation task. The third challenge is capturing the relative importance of different neighbours in the recommendation graph. newline