engleski

On the reciprocal links in complex networks

The abundance of reciprocal edges present one of the simplest type of local structural correlation present in directed graphs. Although a huge amount of work is done in the direction of understanding different types of correlations in complex networks, reciprocity because of its apparent simplicity was more often omitted in the research than not. In this talk the unexpected effects of this simple correlation on other correlation measures will be presented. The previous work regarding reciprocal links showed that different networks can be well classified according to the present number of reciprocal edges (Garlaschelli, 2004). Such classification corroborates the idea that reciprocal edges have significant influence on the function of real networks. The specific role of reciprocal edges is further accessed in the work on correlations in WWW (Serrano, 2005). In the ensemble of growing Wikipedia networks (Zlatic, 2006) is observed the unusual stability of reciprocity measure throughout the Wikipedias written in different languages. Reciprocity is just one of many different measures of network correlations present in recent literature. The problem of assessing the influence of one type of correlations on other types of correlations is still largely open problem in the theoretical network science. Some of the previous works include the influence of degree correlations on clustering coefficient (Dorogovtsev, 2004) and on reciprocity(Lopez, 2008).

complex networks; reciprocal links; direct graphs

nije evidentirano