Most existing recommender systems use a traditional method called Collaborative Filtering (CF) to predict the preferences of target users in order to provide recommendations to help them with their decision making process. CF learns user preferences based on user community’s past behaviors. However, Collaborative Filtering may provide an invalid recommendation if a system does not have explicit feedback or item rating from users. As a result, One-class Collaborative Filtering (OCCF) came into existence since it took only positive examples or implicit feedback into consideration to provide recommendations. With the rapid growing trend of the online social networking services which generate and reflect social relationships among their users as well as consist of representation, interests and activities of each user, there is an opportunity to improve OCCF. Therefore, this research aims to propose a novel method for OCCF using online social relationships to increase a prediction accuracy of the recommendation. It is believed that social relationship data can reflect the social influence as people tend to have a default trust in their friends’ tastes in an online social network. This is also what the CF algorithms try to reveal. In this research, Non-negative Matrix Factorization method was applied with social influence weighting scheme to the one-class problem. Based on the experimental evaluation and two decision-support measures, the proposed method was proved to provide higher quality of recommendation results than the other baseline methods.