There is a substantial amount of current information systems and marketing research focused on social networking sites, most frequently on Facebook. Often, these studies utilize available metadata on user on-line behavior, such as what links the users clicked on. In order to better understand behavior of Facebook users, it makes sense to investigate also whom the users connect to. It is possible to hypothesize that behavior of people, who connect only to relatives on Facebook, differs from behavior of people, who are connected only to their classmates. The paper offers a framework of Facebook connections and a proposal how to group these connections.