An effective user clustering-based collaborative filtering recommender system with grey wolf optimisation

Abstract

The enormous amount of data available today often makes it difficult for users to make decisions. Recommendation systems have become increasingly popular and mainly used in e-commerce to helping predict user preference towards particular items. The proposed system performs user cluster-based collaborative filtering for venue recommendations in which clusters are formed using a bio-inspired grey wolf optimisation algorithm. Clustering is used to eliminate the disadvantages of collaborative filtering regarding scalability, sparsity, and accuracy. In addition, we have used two similarity computation methods, namely the Pearson correlation coefficient (PCC) and cosine similarity to find the similarities between the set of users. The proposed recommendation system with the bio-inspired grey wolf optimisation algorithm has been evaluated on real-world massive volume datasets of Yelp and Trip Advisor for finding out the accuracy, precision, recall, and f-measure. We have also modelled and validated new mobile-based recommendation application frameworks for the development of urban venue recommendations in smart cities. The experimental and evaluation results demonstrate the usefulness of the newly generated recommendations and exhibit user satisfaction with the proposed recommendation technique.