We describe a simple and efficient method - the « Louvain method » - for the detection of communities in networks. The method has sub-quadratic computational complexity and can be routinely used for networks with billions of nodes or links ; the method was chosen a couple of months ago by LinkedIn for its network visualization tool. As an illustration of the method, we describe the communities obtained on a social network constructed from billions of mobile phone communications at country scale. The low complexity of the method allows the analysis of large time-evolving networks and we illustrate this with a vizualisation of how a country-wide social network evolve over a six months period.