An in-depth analysis of Recommender systems for integration of knowledge graphs Utilising federated XAI Model

An in-depth analysis of Recommender systems for integration of knowledge graphs Utilising federated XAI Model

연구 분야: Software Development

학회: International Journal of Information Technology

Recommender systems have emerged as a valuable means of improving user experiences available in a number of modern domains such as e-commerce, entertainment, social networks, and so on in the conditions of information overload. However, these systems suffer from challenges like data sparsity, cold-start problem etc. that hinders its optimal recommendation. As a solution to these challenges, the given paper introduces a federated explainable artificial intelligence (XAI) framework for integration of knowledge graphs (KG’s) in recommender systems. KG’s are characterized by their ability to model entities and links between them showing that they can be used to enrich the semantic information to help understand recommendations and make them more interpretable, scalable and adaptable for users. After capturing user-item interactions, decentralized federated averaging (DFA) concept is applied for updating model parameters followed by prediction of user ratings by using Shapley Additive exPlanations (SHAP). The goal of SHAP is to explain the prediction of an instance by computing the contribution of each feature. Utilization of SHAP as an XAI technique ensures scalability and interoperability of the proposed model. Lastly, the model is validated and compared with existing recent studies based on evaluation metrics such as root mean square error (RMSE) and mean absolute error (MAE).