In the era of the Internet, information overload is a growing problem which refers to the inability of a person to make a decision because the amount of information that she/he needs to process is huge. To solve this problem, recommender systems were proposed to apply various algorithms to recognize users’ preferences and generate recommendations which are likely match the user’s interest on various items. In this thesis, we aim to improve the effectiveness of the recommendation by incorporating the social data into the traditional recommendation algorithms. Hence, we first propose a new user similarity metric that not only considers tagging activities of users, but also incorporates their social relationships, such as friendships and memberships, in measuring the nearest neighbours. Subsequently, we define a new recommendation method which makes use of both user-to-user similarity and item-to-item similarity. Experimental outcomes on a Last.fm dataset show positive results of our proposed approach.