Effector Detection Problem in Social Networks

Nowadays, different innovations spread rapidly in online social networks. An activation state can indicate whether each user adopts the target information. The effector detection problem aims to find a way to generate an activation state as close to an observed one as possible. In this article, base...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on computational social systems Vol. 7; no. 5; pp. 1200 - 1209
Main Authors: Zhang, Yapu, Yang, Wenguo, Wu, Weili, Li, Yi
Format: Journal Article
Language:English
Published: Piscataway IEEE 01.10.2020
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Algorithms
Approximation
Approximation algorithm
Approximation algorithms
Biological system modeling
effector detection
Integrated circuit modeling
Linear programming
Lower bounds
Mathematical analysis
reverse influence sampling (RIS)
Social network services
Social networks
Target detection
Technological innovation
Upper bounds
ISSN:2329-924X, 2373-7476
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Nowadays, different innovations spread rapidly in online social networks. An activation state can indicate whether each user adopts the target information. The effector detection problem aims to find a way to generate an activation state as close to an observed one as possible. In this article, based on the influence spread, the unconstrained and constrained effector detection problems are proposed. To tackle them, we design two approximation algorithms since the problem is NP-hard, and the objective function is nonsubmodular. For the unconstrained case, our objective function can be best provided with the difference of two submodular functions. Thus, we address this problem through the modular-modular algorithm. For the constrained case, we devise the solutions for the original function, submodular upper bound, and lower bound according to an idea of reverse influence sampling. Then, there is a data-dependent approximate solution using the sandwich approximation algorithm. Finally, we show the correctness and superiority of our methods through massive experiments in three real-world networks.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2329-924X
2373-7476
DOI:10.1109/TCSS.2020.3013734