Community Dtection Algorithm Reasearch Based On Density Clustering

Network is an effective way to present complex systems.We define the final form that contains a large number of nodes and the edges connected by these nodes as complex networks.The large-scale complex networks(such as social networks,transportation systems,engineering infrastructures and so on)have shown some non-trivial universal structural properties compared to regular networks.Depending on the nature,complex networks can be divided into clusters or modules of nodes and edges,which can also be called communities.Complex network clustering can help scholars to develop complex network structures,study the inherent attributes of complex networks and predict complex network trends and other aspects of the work.Density-based clustering algorithms can find irregularly shaped clusters and cluster data without knowing in advance of the number of clusters they contains,So it is more suitable for complex networks in the real world.SCAN algorithm that is a typical representative of the density clustering algorithm but most of them have a high time complexity and does not apply to large-scale complex networks.The aim of this paper is to propose the QSCAN algorithm based on SCAN density clustering algorithms,which can deal with the community structure for large scale complex networks.In the sake of reducing the time complexity,we introduce a new data set D,which is a two-level node set with a given node.The QSCAN algorithms can maintain the clustering efficiency by using the following two methods.On the one hand,in order to reduce the time complexity of the algorithm,QSCAN only calculates the density of adjacent nodes in the data set D.on the other hand,through the part of the data set D known node similarity,it can also effectively calculate the last data set node similarity.Therefore,the QSCAN algorithm can detect the same clustering under the condition of saving time,after the clustering is completed,the QSCAN algorithm reclassifies the remaining nodes that do not belong to this cluster as the hub and outlier.