Influencer Selection on Social networks based on Information Requirement and Diffusion Cost
Olanrewaju Abdus-Samad Temitope1, Ahmad Rahayu2, Massudi Mahmudin3
1Olanrewaju Abdus-Samad Temitope, School of Computing, University Utara Malaysia Sintok, Kedah, Malaysia.
2Ahmad Rahayu, School of Computing, University Utara Malaysia Sintok, Kedah, Malaysia.
3Massudi Mahmudin, School of Computing, University Utara Malaysia Sintok, Kedah, Malaysia.
Manuscript received on 03 February 2019 | Revised Manuscript received on 10 February 2019 | Manuscript Published on 22 March 2019 | PP: 221-235 | Volume-8 Issue-5S April 2019 | Retrieval Number: ES3420018319/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Viral marketing is vital to the success of business in this age. Information diffusion on social networks for viral marketing involves selecting a seed set of influencers (nodes) to be infected which leads to an activation process in the network with the aim of infecting a maximum number of nodes. The existing models have selected the influencers based on the node properties (centralities) but do not take into consideration the diffusion cost in spreading the information. In addition, the influencers are selected without considering the need for diffusing information. This study proposes a general additive model that uses a tuneable weight on four centralities in selecting influencers. Our results shed more light on the trade-off between the outreach of information and the diffusion cost incurred. The results demonstrated that selecting the top influencers using a single metrics is not necessarily effective when diffusing information. This study also discovered a positive effect in an increase of the size of the influencers does not always yield an increase in the relative outreach depending on the type of the network.
Keywords: Social Networks, Centrality, Information Diffusion, Diffusion Cost.
Scope of the Article: Evolutionary Computing and Intelligent Systems