Roadmap/Future Work#
There are a few areas of development we’d like to prioritize for the future of cosasi:
More localization algorithms. The source inference space is fairly large. We’re always looking to extend existing functionality to cover a greater portion of the algorithms prominent in literature.
Handle more complex contagions. We currently implement graph diffusion on static networks. While this can be a satisfactory approximation, realistic contact networks can vary over time; we would like to handle this situation, as well.
Implement tools for running and detecting adversarial methods. Recent progress has been made toward protocols that obfuscate sources, although standard methods like adaptive diffusion are not resilient to adversaries with access to three or more independent “snapshots” (source). We would like to extend
cosasito handle both sides of this problem.