An Approach Of Graph Ranking Based On Diversity Of Nodes

For central-based graph ranking,which is one of the most important graph ranking,existing many applications in real-life scenarios such as advertisement recommendation,commodity marketing and webpage ranking.Centrality is the measurement of the importance of the nodes in the graph.According to different definitions of importance,there are three major measures,respectively degree centrality,betweenness centrality and closeness centrality.The diversity which we focus on can be considered as a kind of centrality,it distinguishes the roles of nodes in network by the numbers of neighbors and the difference between neighbors.However,with the rapid development of online community and network media,graph datas are becoming very large and often have billions of vertices and trillions of edges,together with complex data relationships,which makes the original algorithm of graph ranking based on diversity of nodes not suitable.Moreover,the original method of mining diversity of nodes is completely based on the neighbor topology,do not consider the nodes attributes.Nowadays with rich information available from graph data,which nodes are associated with a set of attributes describing their properties.Node attributes should be used for capturing the diversity of nodes in attributed graph data.In this paper,starting from a great deal of graph data,through the introduction of a new computing system of graph named GraphX that is a component in Spark for graph-parallel computation,we firstly improve original graph ranking based on diversity of nodes algorithm in the non-attributed graph.And then,we propose graph ranking based on diversity of nodes algorithm in the attributed graph.Some works can be summarized as follows:(1)Base on the idea of attribute coverage,defining the metric function of the diversity of nodes in the attributed graph.(2)By the distributed graph processing framework GraphX of Spark,programming and implementing graph ranking based on diversity of nodes algorithm included attributed graph and non-attributed graph.(3)Testing all the involved methods in the real graph data sets.Then,analyzing the experiment results and give out the conclusion.