| In our daily life,the connection among things constitutes many networks.As the connection becomes more and more complex,complex networks have attracted extensive interest and research in which the influence maximization becomes a hot issue.The greedy-based algorithms often achieve higher influence spread,but they are timeconsuming.Compared with the greedy-based algorithms,the heuristic-based algorithms are efficient.However,the algorithms do not consider the network structure and the intrinsic relationship between the nodes so that the influence spread of them is not ideal.The heuristic-based algorithms and the greedy-based algorithms are difficult to meet the requirements of accuracy and efficiency,simultaneously.Therefore,this paper proposed an algorithm(MaxCliDN)based on the attenuation of degree and the maximum clique of the network,which reduces the influence coverage between adjacent nodes by the maximum clique of the network for expanding the influence spread of the seed nodes that are selected by the heuristic-based algorithms.At the same time,for solving the problem of high computational complexity of the greedy-based algorithms,this paper proposed a node influence ranking algorithm(Deg_Ncliq),which uses the sum of the current node's degree and the number of communities in the neighboring nodes(D_Ncilque)as the index to select seed nodes,and it can reduce the running time of the CELF algorithm.In order to verify the effectiveness and efficiency of the proposed algorithms,the two proposed algorithms were compared with the existing algorithms in seven public datasets with the independent cascade model.The experiment results show that MaxCliDN is outperform the traditional heuristic-based algorithms such as the degreecentrality,the clossness-centrality and the betweenness-centrality algorithm in terms of influence spread.At the same time,the proposed Deg_Ncliq achieves better running time comparing with the traditional CELF algorithm with matching influence spread. |
PDF Full Text Request