Proposed Recommender System For Solving Cold Start Issue Using k-means Clustering and Reinforcement Learning Agent

Abstract

The cold start problem for new users is making a real challenge in the recommender system's operation to provide suggestions for a new user. This paper suggests a reinforcement learning recommender system that provides the automatic systems of multi-armed bandits which are learning and improving its efficiency from experience without being explicitly programmed. So, a movie recommender system is built using the K-Means Clustering to cluster the dataset and epsilon greedy reinforcement learning agent to manage multi-armed bandits recommendation process. The recommender system consists of multi-armed bandits that connect to five clustered datasets represents five movie genres and that was the first contribution. The second contribution is checking whether the NDCG is sufficient in measuring the quality of services in multi-armed bandits' recommender systems. The proposed recommender system has been tested using a movie lens 100-K dataset. The system measured using accumulative gain, RMSE, and NDCG. The results are showing efficiency to learn new user preferences.