A noise correction-based approach to support a recommender system in a highly sparse rating environment

Abstract

Recommender systems support consumers in decision-making for selecting desired products or services from an overloaded search space. However, this decision support system faces difficulties while dealing with sparse and noisy rating data. Therefore, this research re-classifies users and items of a system into three classes, namely strong, average and weak to identify and correct noise ratings. Later, the Bhattacharya coefficient, a well-performing similarity measure for a sparse dataset, is integrated with the proposed re-classification method to predict unrated items from the obtained noise-free sparse dataset and recommend preferred products to consumers. Furthermore, the effectiveness of the proposed model is validated on two sparse and noisy datasets and compared with various published methods in terms of the mean absolute error (MAE), root mean square error (RMSE), F1-measure, precision, and recall values. The obtained results confirm that the proposed model performs better than other published relevant methods.