| Abstract |
Recently, the availability of large datasets on social networks has significantly facilitated the study of influence prop- agation, providing real-world benchmarks for validating theoretical models. Although this has become a prominent topic of research, little work analyses influence propagation under information constraints (i.e., partial and/or uncer- tain information). However, such constraints are common in real-world settings, and especially in coalition operations, where parts of an external group may be intentionally obscured by adversaries and where its structure must be inferred from noisy surveillance data. We propose new approaches that exploit generative models of network topology to significantly improve influence diffusion in partially observable networks. We show that these algorithms perform similarly to or better than the state-of-the-art, showing promise for future research in this direction. |