Effectiveness of Graph Neural Networks for User-User-Item Recommendation Systems

Abstract

Graph Neural Networks are an efficient and effective framework for graph representation learning. The versatility and representational power of graph neural networks are quickly making them the preferred tool for many machine learning tasks. The same is true of recommender systems: graph neural networks allow us to easily incorporate new user, item, and contextual features. Recommender systems take a set of existing features as input to get embedding of user's. Standard high-performance deep architectures assume particular data symmetries which are not satisfied in recommender systems. So, this research work has proposed a Graph neural network (GNN) based recommendation system, where interactions of user-item are taken into consideration. Both interactions and opinions are encoded to build a user-item graph in the proposed approach. To obtain strong recommendations trust information has an important role. Also, this research work has developed a trust circle to get the nearest neighboring users for item recommendations in order to take social relations into consideration. Extensive experiments on real-world datasets have been conducted to demonstrate the superiority of the proposed model and it significantly outperforms baseline methods.