TY - GEN
T1 - DARIM
T2 - 26th International Conference on Neural Information Processing, ICONIP 2019
AU - Hosni, Adil Imad Eddine
AU - Li, Kan
AU - Ahmad, Sadique
N1 - Publisher Copyright:
© 2019, Springer Nature Switzerland AG.
PY - 2019
Y1 - 2019
N2 - This paper investigates the problem of rumor influence minimization in online social networks (OSNs). Over the years, researchers have proposed strategies to diminish the influence of rumor mainly divided into two well-known methods, namely the anti-rumor campaign strategy and the blocking nodes strategy. Although these strategies have proven to be efficient in different scenarios, their gaps remain in other situations. Therefore, we introduce in this work the dynamic approach for rumor influence minimization (DARIM) that aims to overcome these shortcomings and exploit their advantage. The objective is to find a compromise between the blocking nodes and anti-rumor campaign strategies that minimize the most the influence of a rumor. Accordingly, we present a solution formulated from the perspective of a network inference problem by exploiting the survival theory. Thus, we introduce a greedy algorithm based on the likelihood principle. Since the problem is NP-hard, we prove the objective function is submodular and monotone and provide an approximation within $$(1-1/\textit{e})$$ of the optimal solution. Experiments performed in real multiplex and single OSNs provide evidence about the performance of the proposed algorithm compared the work of literature.
AB - This paper investigates the problem of rumor influence minimization in online social networks (OSNs). Over the years, researchers have proposed strategies to diminish the influence of rumor mainly divided into two well-known methods, namely the anti-rumor campaign strategy and the blocking nodes strategy. Although these strategies have proven to be efficient in different scenarios, their gaps remain in other situations. Therefore, we introduce in this work the dynamic approach for rumor influence minimization (DARIM) that aims to overcome these shortcomings and exploit their advantage. The objective is to find a compromise between the blocking nodes and anti-rumor campaign strategies that minimize the most the influence of a rumor. Accordingly, we present a solution formulated from the perspective of a network inference problem by exploiting the survival theory. Thus, we introduce a greedy algorithm based on the likelihood principle. Since the problem is NP-hard, we prove the objective function is submodular and monotone and provide an approximation within $$(1-1/\textit{e})$$ of the optimal solution. Experiments performed in real multiplex and single OSNs provide evidence about the performance of the proposed algorithm compared the work of literature.
KW - Anti-rumor campaign strategy
KW - Blocking nodes strategy
KW - Rumor influence minimization
KW - Rumor propagation
UR - http://www.scopus.com/inward/record.url?scp=85076931333&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-36711-4_52
DO - 10.1007/978-3-030-36711-4_52
M3 - Conference contribution
AN - SCOPUS:85076931333
SN - 9783030367107
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 619
EP - 630
BT - Neural Information Processing - 26th International Conference, ICONIP 2019, Proceedings
A2 - Gedeon, Tom
A2 - Wong, Kok Wai
A2 - Lee, Minho
PB - Springer
Y2 - 12 December 2019 through 15 December 2019
ER -