Competitive Influence Maximization: Countering Disinformation with Algorithms
How can we best fight influence campaigns of trolls and bots? Competitive influence maximization uses graph algorithms and the structure of social networks to defeat the attackers.
In competitive influence maximization, we are given a network. The attacker selects vertices to influence, and the defender reacts by selecting other vertices to oppose the spread of the attacker's message. The problem is too hard to solve optimally. The goal is to design an algorithm for the defender that can cope with uncertainty and find an approximate best response to any move made by the attacker.
Goal
The goal of this project is to implement a best response algorithm for competitive influence maximization using C/C++. A possible extension is to parallelize the algorithm and/or port it to GPUs using CUDA.
Learning outcome
Efficient implementation of graph algorithms in C/C++. Enhanced understanding of network science. Understanding of formal models for influence operations.
Qualifications
Good understanding of graph algorithms, C/C++ programing, parallel programing in CUDA/OpenMP is optional
Supervisors
- Johannes Langguth
Collaboration partners
- University of Bergen
- United State Military Academy West Point
References
- Kempe, D., Kleinberg, J., & Tardos, É. (2003). Maximizing the spread of influence through a social network. (https://dl.acm.org/doi/abs/10.1145/956750.956769)