In the last few years, there has been a great interest in detecting overlapping communities in complex networks, which is understood as dense groups of nodes featuring a low outbound density. To date, most methods used to compute such communities stem from the field of disjoint community detection by either extending the concept of modularity to an overlapping context or by attempting to decompose the whole set of nodes into several possibly overlapping subsets. In this report we take an orthogonal approach by introducing a metric, the cohesion, rooted in sociological considerations. The cohesion quantifies the community-ness of one given set of nodes, based on the notions of triangles - triplets of connected nodes - and weak ties, instead of the classical view using only edge density. A set of nodes has a high cohesion if it features a high density of triangles and intersects few triangles with the rest of the network. As such, we introduce a numerical characterization of communities: sets of nodes featuring a high cohesion. We then present a new approach to the problem of overlapping communities by introducing the concept of ego-munities, which are subjective communities centered around a given node, specifically inside its neighborhood. We build upon the cohesion to construct a heuristic algorithm which outputs a node's ego-munities by attempting to maximize their cohesion. We illustrate the pertinence of our method with a detailed description of one person's ego-munities among Facebook friends. We finally conclude by describing promising applications of ego-munities such as information inference and interest recommendations, and present a possible extension to cohesion in the case of weighted networks.