High-throughput techniques investigating for example protein-protein or protein-ligand interactions produce vast quantity of data, which can conveniently be represented in form of matrices and can as a whole be regarded as knowledge networks. Such large networks can inherently contain more information on the system under study than is explicit from the data itself. Two different algorithms have previously been developed for economical and social problems to extract such hidden information. Based on three different examples from the field of proteomics and genetic networks, we demonstrate the great potential of applying these algorithms to a variety of biological problems.