Measuring Quadrangle Formation in Complex Networks

The classic clustering coefficient and the lately proposed closure quantify formation of triangles from two different perspectives, with focal node at centre or end in an open triad respectively. As many networks are naturally rich triangles, they become standard metrics to describe analyse networks. However, advantages applying them can be limited networks, where there relatively few but which quadrangles, such as protein-protein interaction neural food webs. This yields for other approaches that would leverage quadrangles our journey better understand local structures their meaning types Here we propose quadrangle coefficients, i.e., i-quad o-quad coefficient, further extend weighted Through experiments on 16 six domains, first reveal density distribution then correlations degree. Finally, demonstrate network-level, adding average leads significant improvement network classification, while node-level, coefficients useful features improve link prediction.