Social networking sites have experienced an explosion in both the number of users and the amount of shared information in recent years. Thanks to the positioning function in mobile devices, e.g., GPS, location-based social networks (LBSNs) become a prominent representative of social networks. Thus, the rapid development of LBSN services has stimulated the emergence of a new line of research to develop two novel recommender systems that seek to facilitate users' location exploration and social interaction, where we refer them to location and friend recommender systems, respectively. Even with the help of large available user interaction data, it is challenging to produce accurate location recommendation and friend recommendation in LBSNs by the reason of the interdependency between human mobility and social proximity, and the heterogeneity of social link, i.e. connection between user and user, and consumption link, i.e. connection from user to location, in the whole network. To this end, in this dissertation, we propose different approaches for these two types of recommender systems with the characteristics observed in user behaviors to serve LBSNs. For location recommendation, we study human mobility and exploit human movement properties to design recommender systems in spatial, temporal, and social aspects for helping users discover the desired locations. For friend recommendation, a new approach is introduced to maximize the growth of both heterogeneous links in the whole network for helping users find potential friends to connect with. The performance evaluated on real-world datasets demonstrates the ability of our models for recommender systems in LBSNs.