Links between datasets are central for manifold applications on the Web of Data, including federated queries and cross-ontology question answering. Yet, establishing links between datasets is a tedious problem w.r.t. its accuracy and runtime. In this talk, we present LIMES, a scalable and accurate Link Discovery framework for the Web of Data. We present the architecture of the tool as well as some of the algorithms it contains. Moreover, we evaluate the current state of the art and present some of the challenges currently faced in the Link Discovery area.