Winning hearts and minds: Maximizing influence in social networks
Military / Coalition Issue
When military efforts include civilian engagement (e.g. crowd-sourcing activities), maximum participation, especially in the presence of an adversary relies on tactical interaction with the public. Here the main challenge is the identification and effective incentivisation of key individuals in a social network who can then spread influence in the rest of the population. The issue is particularly relevant in coalition settings, where competition may arise between allies as they compete to recruit human agents (or soft sensors) in a network to accomplish self-agendas or to have greater control over a joint operation.
Core idea and key achievements
The problem is widely explored in several real-world scenarios where in each case we present theoretical solutions that exploit network topology to determine optimal allocation of limited resources to counter adversarial influence in social networks. More specifically, we characterise optimal strategies under various constraints, such as network uncertainty and presence of negative (or antagonistic) ties. We also examine the impact of propagation errors and the effect of constrained access to nodes in a network. In every case, we derive optimal solutions both under complete knowledge of the adversary’s strategy and under game-theoretic settings without any information about the adversary. Lastly, we design a human-subject experiment to establish if people employ rational or inherently biased strategies when maximising opinion spread in the real-world context.
Implications for Defence
The proposed algorithms can be readily used to offer real-time solutions to the competitive influence maximisation problem in several defence scenarios (for example, penetration of local networks for knowledge and support in combat zones). The work can also help identify vulnerabilities in any given network that can further aid the mitigation of misinformation spread. Finally, it can serve as a foundation for empirical analysis of online and offline behaviour within several defence (and security) operations.
Readiness & alternative Defence uses
The work is currently at technology readiness level (TRL) 3. An alternative use of the game in the human-subject experiment could be to exploit it as a training tool for military personnel.
Resources and references
- Chakraborty, Sukankana, Sebastian Stein, Markus Brede, Ananthram Swami, Geeth de Mel, and Valerio Restocchi. “Competitive influence maximisation using voting dynamics.” In Proceedings of the 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, pp. 978-985. 2019.
- Brede, Markus, Valerio Restocchi, and Sebastian Stein. “Effects of time horizons on influence maximization in the voter dynamics.” Journal of Complex Networks 7, no. 3 (2019): 445-468.
- Eshghi, Soheil, Setareh Maghsudi, Valerio Restocchi, Sebastian Stein, and Leandros Tassiulas. “Efficient influence maximization under network uncertainty.” In IEEE INFOCOM 2019-IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS), pp. 365-371. IEEE, 2019.
- Brede, Markus, Valerio Restocchi, and Sebastian Stein. “Transmission errors and influence maximization in the voter model.” Journal of Statistical Mechanics: Theory and Experiment 2019, no. 3 (2019): 033401.
- Brede, Markus, Valerio Restocchi, and Sebastian Stein. “Resisting influence: how the strength of predispositions to resist control can change strategies for optimal opinion control in the voter model.” Frontiers in Robotics and AI 5 (2018): 34.
- Stein, Sebastian, Soheil Eshghi, Setareh Maghsudi, Leandros Tassiulas, Rachel KE Bellamy, and Nicholas R. Jennings. “Heuristic algorithms for influence maximization in partially observable social networks.” In SocInf@ IJCAI. 2017.
University of Southampton, IBM (UK and US), ARL.