Optimized algorithm for skyline community discovery in multi-valued networks

Abstract

The community discovery problem aims to find the close subgraphs from the traditional network. The traditional network with multiple numerical attributes of nodes is called the multi-valued network. The Skyline Community is the largest connected k-core that is not dominated in a multi-valued network, which can be used to solve multi-objective decision-making problems. In this paper, we focus on the skyline community discovery problem. Firstly, we propose a combined index that includes the degree-neighbor (D-N) index and inverted index to manage the multi-valued network. The index can filter the nodes whose degrees do not satisfy k. Simultaneously, it can accelerate the judgment of the neighbor relationship between the nodes and improve the query efficiency. Then, we propose the SCDCI (Skyline Community Discovery Algorithm Based on Combined Index Structure) algorithm to compute the skyline community. The algorithm scans the nodes in turn from the inverted indexes, avoiding lots of redundant calculations. The nodes are scanned in descending order by their attributes, therefore, lots of nodes that cannot form a skyline community do not be processed. Finally, the effectiveness of SCDCI algorithm is verified through experiments. Compared with the existing algorithms, the efficiency of our algorithm is improved a lot.