Development of a recommender system based on a trust approach in social networks using the K-NN algorithm

Abstract

In recent years, the increase in the volume of information has made it difficult to make accurate decisions about a specific matter. Recommender systems have been proposed as a suitable solution to handle this challenge. These systems use the processing of user comments and provide only useful information to them. This study aims to provide a trust-based recommender system in social networks using the K-nearest neighbor algorithm. Several operational tests have been performed on five datasets taken from GroupLens. Analysis of the results showed that the MAE value of the proposed method is better than the other three compared methods. This improvement is evident in all results from five data sets as well as the mean percentage of results (proposed method: 92.28, Bayesian: 84.45, CBR: 82.28, and RBF: 90.84). The average results show that the MAE of the proposed method is better compared to other methods. In general, the proposed method has better accuracy based on MAE than the other three methods.