Latent relationship discovery, such as the recommendation of potential friends, is a common feature in many social networking sites such as Facebook. In an enterprise, discovery of latent relationships is particularly important in helping knowledge workers discover others in their company with specific expertise or shared interests. This paper describes a methodology that makes use of user‐generated content and incorporates multiple dimensions, such as social distance, semantic distance, geographic distance, temporal distance, and quantity, to determine the strength of latent social relationships. Examples from the People & Projects enterprise social networking application and from work exploring the visualization of conversation patterns are used to illustrate the manner in which these latent relationships can be intuitively presented to users. Beyond social networking applications, this methodology has direct applicability in other venues such as contact centers, where the discovery of latent relationships between customers, agents, and contact sessions can be used to improve service and performance. © 2012 Alcatel‐Lucent.