| Multi-Round Influence Maximization(MRIM)is an important research content in the field of social network analysis,and it has good application value in marketing and advertising.The MRIM algorithm is divided into two types: adaptive and non-adaptive.The non-adaptive MRIM algorithm generally determines the set of seed nodes for all rounds in advance through a central index or a greedy strategy,which has high efficiency but poor accuracy.The adaptive MRIM algorithm can adaptively select the seed node of the current round based on the feedback of the active node in the previous round,so as to ensure the accuracy of seed selection.However,this type of algorithm is mostly based on a greedy strategy and has low efficiency.This thesis studies the above-mentioned problems of the adaptive multi-round influence maximization algorithm,and designs the seed node selection methods for the first round of propagation and the remaining round of propagation,which not only ensures the accuracy of seed node selection,but also takes into account the efficiency.main tasks as follows:(1)The first-round influence maximization algorithm based on coverage and structural holesTraditional topology-based influence maximization algorithms select seed nodes with greater influence through centrality indicators.However,this type of algorithm often only considers a single attribute of the network,and the selected centrality indicators are difficult to adapt to networks with different structural characteristics,leading to unstable performance.This thesis proposes an influence maximization algorithm(NCSH)based on coverage and structural holes from the perspective of multi-attribute fusion.The algorithm combines the advantages of the number of neighbors covered by the node and the location advantage,and effectively solves the problem of unstable performance of the traditional topology-based influence maximization algorithm.(2)Adaptive multi-round influence maximization algorithm based on improved degree discountTraditional adaptive multi-round influence maximization algorithms are mostly based on greedy strategies.Choosing a seed node needs to execute Monte Carlo simulation propagation thousands of times,which causes the algorithm to run too long.Starting from the network topology,this thesis proposes an adaptive multi-round maximization algorithm based on improvement degree discount(NCSH＿IDD),which integrates the first-round influence based on coverage and structural holes proposed in this paper.Based on the idea of degree discount,the calculation method of the influence of nodes in the remaining round propagation is improved,and the time complexity of the algorithm is effectively reduced.In this thesis,there are 10 pictures,2 tables,and 59 references. |
