This paper introduces a novel approach to identifying communities in networks by focusing on links rather than nodes. Traditional community detection methods often fail to capture the overlapping nature of communities and the hierarchical structure of networks. By considering communities as groups of links rather than nodes, the authors demonstrate that this approach effectively reconciles the conflicting principles of overlapping communities and hierarchy. Link communities naturally incorporate overlap while revealing hierarchical organization, and are found in many real-world networks, including biological and social networks. The authors propose a method for identifying link communities using hierarchical clustering based on link similarity. This method allows for the simultaneous detection of hierarchical and overlapping relationships. They define a measure called partition density, D, which quantifies the quality of link communities and is used to determine the optimal level for cutting the dendrogram to extract communities. The study shows that link communities reveal more about the metadata of networks than traditional node-based methods. The authors compare their method with three widely used community detection algorithms: clique percolation, greedy modularity optimization, and Infomap. Their results show that link communities outperform these methods in terms of community quality, coverage, and overlap. The study also highlights the importance of considering link-based approaches for understanding complex networks, especially those with high levels of overlap. The authors provide evidence that link communities are fundamental building blocks that reveal overlap and hierarchical organization as two aspects of the same phenomenon. The results suggest that link-based approaches are superior to node-based approaches in capturing the structure and dynamics of complex networks. The study also provides quantitative evidence that link communities reveal more about the metadata of networks than traditional node-based methods. The authors conclude that link-based approaches are essential for understanding complex networks, particularly those with high levels of overlap.This paper introduces a novel approach to identifying communities in networks by focusing on links rather than nodes. Traditional community detection methods often fail to capture the overlapping nature of communities and the hierarchical structure of networks. By considering communities as groups of links rather than nodes, the authors demonstrate that this approach effectively reconciles the conflicting principles of overlapping communities and hierarchy. Link communities naturally incorporate overlap while revealing hierarchical organization, and are found in many real-world networks, including biological and social networks. The authors propose a method for identifying link communities using hierarchical clustering based on link similarity. This method allows for the simultaneous detection of hierarchical and overlapping relationships. They define a measure called partition density, D, which quantifies the quality of link communities and is used to determine the optimal level for cutting the dendrogram to extract communities. The study shows that link communities reveal more about the metadata of networks than traditional node-based methods. The authors compare their method with three widely used community detection algorithms: clique percolation, greedy modularity optimization, and Infomap. Their results show that link communities outperform these methods in terms of community quality, coverage, and overlap. The study also highlights the importance of considering link-based approaches for understanding complex networks, especially those with high levels of overlap. The authors provide evidence that link communities are fundamental building blocks that reveal overlap and hierarchical organization as two aspects of the same phenomenon. The results suggest that link-based approaches are superior to node-based approaches in capturing the structure and dynamics of complex networks. The study also provides quantitative evidence that link communities reveal more about the metadata of networks than traditional node-based methods. The authors conclude that link-based approaches are essential for understanding complex networks, particularly those with high levels of overlap.