| In real life as well as research work applied to the various units and the relationships between them can be abstracted into a network, due to the complexity of the network information, call such networks complex networks. Structure of complex networks is a complex system, including the structural complexity: is the network system has a rich structure he includes community, motif, clustering, generate regularity. Structure of the network may change over time; the complexity of the node, which includes the complex interaction between network complexity and complexity of the network hierarchy; network evolution, and in the node or link generation and disappear, this also shows that the time-varying network structure; junctional diversity, including his connection weights and diversity diverse directions; kinetics complexity and multiple and complex fusion and so on. All of the above features suggest that the complexity of the study can be generalized to discuss the many ways networks.Complex network nodes are distributed according to gather community feature, it can be divided into single sub-network and bipartite network. Complex networks exist between all nodes are connected relationship or potential relationship network connection points called single network; however bipartite network is all the nodes are divided into two sets, there is no connection between the two sets of internal, connection relationship exists between the collection or for possible connection relationship.Prediction network link means through the existing node connection relationship to predict the possibility of connection between the nodes connection relationship does not exist. This forecast includes both the link prediction of would not exist for itself and does not exist in the future, but also contains a link prediction of exist forecast in the future. Work done as follows:First, understand the complexity of networks and knowledge network links predictable, find bipartite network implementation in complex networks of reciprocity and hunting behavior through biological populations between networks, also found that there are still many signs of bipartite network in real life. Through the bipartite network characteristics to understand, to find the link on bipartite network prediction method unique, it is not limited to the existing general link prediction. This link prediction method is based on the idea the local path algorithm. First observed bipartite network path lengths there is only the odd path, from the exponential function to associate metric path can be obtained even after deleting an odd path, and this odd path from a mathematical point of view, the formula is the trigonometric hyperbolic sine function; also include von Neumann indicators were equally odd part of the reservation to link predict bipartite network.Through the bipartite network of understanding and forecasting, conducting path matrix analysis found that there is a problem repeated path in the path matrix, and the path length of the longer, the more the number of repeated, resulting in unnecessary waste of resources,and to a certain extent affect the observation and understanding of the true path information of the network. Delete repeat path problem has become the main issue discussed at the present stage of this article. Through the channel matrix generated in the form of observation, find out why repeat path generated and delete duplicate path approach.After removing duplicate path prediction could well get the desired results.Association from bipartite network to the general network whether there actually also exist repeat path. The answer is yes. However, due to the natural of the network it does not distinguish between odd-path and even-path. Adoption with the same thinking of bipartite network to remove repeat path, then conducted the experimental analysis. |
