Font Size: a A A

Study On The Commensal Influence Maximization Of Social Networks

Posted on:2024-08-11Degree:MasterType:Thesis
Country:ChinaCandidate:T TongFull Text:PDF
GTID:2530307118473004Subject:Software engineering
Abstract/Summary:
Influence maximization is an important research in the field of social network analysis.In response to the phenomenon that many giant companies open multiple subsidiaries and operate different grades of products for different groups and audiences,commensal influence maximization focuses on how to coordinate the seed selection strategies among subsidiaries to achieve the purpose of maximizing the influence of the parent company.For the single-round commensal influence maximization problem,this thesis proposes Single-Round Commensal Influence Maximization Algorithm(SR-CIM),which consists of two parts:Two-Hop Incremental Centrality Algorithm for Preliminary Selection(2-Hop ICA)and Hybrid Matrix Strategy for Overlapping Seed Allocation(HMS).2-Hop ICA strategy is used to perform the initial seed selection for the subsidiaries,and for the overlapping seed problem arising between subsidiaries,HMS strategy is used for redistribution,and the final seed set of each subsidiary is obtained through this process.In this thesis,the control variates is used to verify the effectiveness of the SR-CIM partial and overall algorithm through three experiments.The experiments show that in the single-round commensal influence maximization problem,the SR-CIM algorithm in this thesis does ensure the maximization of the overall influence of the parent company.For the multi-round commensal influence maximization problem,this thesis proposes Multi-Round Commensal Influence Maximization Algorithm(MR-CIM)based on the reality that a company’s promotional products will be promoted multiple times.The algorithm consists of two parts: Two-Hop Incremental Centrality Algorithm with Public Neighbor Loss Coefficient(2-Hop ICA-PNLC)for preliminary seed selection in each round,and the overlapping seed allocation stratagy based on Hybrid Matrix with “Trust Factor”(HMS-TF)is used to redistribute the overlapping seeds generated in multiple rounds.After experimental analysis and multiple comparisons,MR-CIM algorithm in this thesis does effectively improve the overall influence range of the parent company in a multi-round environment.
Keywords/Search Tags:commensal influence maximization, single-round, multi-round, two-hop incremental, hybrid matrix
Related items