Link recommendation, which suggests links to connect currently unlinked users, is a key functionality offered by major online social networking platforms. Salient examples of link recommendation include "people you may know"' on Facebook and "who to follow" on Twitter. A social networking platform has two types of stakeholder: users who use the platform to make and communicate with friends; and operators who establish and operate the platform for commercial benefits. Therefore, the potential advantages of link recommendation are twofold: assisting users to find friends thus improving their usage experience and increasing the benefit (e.g., revenue) of operators through a more connected social graph among users. Current link recommendation methods recommend links based on the likelihood of linkage but ignore the benefit of linkage. That is, these methods recommend links that are more likely to be established in the future but neglect the benefit a recommended link could bring to operators. Such methods thus are designed for the experience of users rather than the benefit of operators. In this study, I propose to recommend links to accommodate the interests of both users and operators. The benefit of linkage is defined using a utility function and the utility-based link recommendation problem are introduced: how to recommend links that are most likely to be established by users and beneficial to operators. To address the problem, I identify key factors underlying utility-based link recommendation decisions based upon relevant social network theories, and construct a Bayesian network model with latent variable accordingly. The optimal parameters of the Bayesian network model are then learned using appropriate machine learning theories and techniques. Analytical properties of the proposed learning method are provided in terms of convergence and close-form updating. Using data obtained from a major social networking platform in the US, evaluations are carefully designed to empirically demonstrate the effectiveness of the proposed utility-based link recommendation method. Evaluation results corroborate that the link recommendation problem is not merely a trivial extension of link prediction tasks. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]