Chathura’s doctoral dissertation, Models of Information Diffusion and the Role of
Influence, examined how local behavioral rules and network structure combine to
produce large-scale patterns of information spread. That question — how does
individual behavior aggregate into population-level diffusion dynamics? — anchors this
entire line of work.
From diffusion to influence
Diffusion models describe that information spreads; influence measurement asks
which actors and relationships actually drove that spread. Moving from one to the
other requires methods that can attribute directional influence between individuals
and communities, not just describe aggregate spread curves.
Measuring pathways
Through the DARPA Modeling Influence Pathways (MIPS) initiative — a sustained,
cross-institutional collaboration between UCF, North Carolina State University, and
MIT — Chathura developed methodology and an interactive dashboard using
transfer-entropy-based methods to identify inter- and intra-community influence
pathways in information-diffusion environments.
Transfer entropy
Transfer entropy is an information-theoretic measure of directed influence between
time series: it quantifies how much knowing one actor’s past behavior reduces
uncertainty about another actor’s future behavior, beyond what that actor’s own
history already explains. Applied to social media activity, it lets researchers infer
directional influence pathways without assuming a fixed network structure in advance.
Influence-network construction
To support this kind of analysis at scale, Chathura developed the Influence Network
Generator (ING), an open-source Python package for constructing influence networks
from online social media data, later adopted in further DARPA-funded diffusion
research.
Interactive analysis
Chathura conceived the methodology and led algorithm design, system architecture, and
visualization for the Transfer Entropy Interactive Dashboard, which was presented
as the official final deliverable to the DARPA-MIPS program.
My contribution
Chathura’s dissertation work established the diffusion-modeling foundation for this
project; he was sole developer of ING and lead developer of the Transfer Entropy
Interactive Dashboard, and is an author across the diffusion and influence
publications listed below.
This project’s influence-measurement methods build on the behavioral premises first
developed for the Multi-Action Cascade Model in the DARPA SocialSim work, and its
statistical treatment of user behavior connects to the categorical functional data
analysis project.