Pranav Pamidighantam's Preliminary Exam
PhD student Pranav Pamidighantam will present his dissertation proposal, "Predicting Influence and Susceptibility of Actors Using Community-Based Approaches in Large-Scale, Attributed, and Dynamic Networks." His preliminary examination committee includes Professor Eunice E. Santos (Chair), Professor J. Stephen Downie, Teaching Associate Professor Chrysafis Vogiatzis, and Assistant Professor Jiaqi Ma.
Abstract
As large-scale networked data continues to become ubiquitous, developing computationally efficient methods for analyzing various processes on these networks is an increasingly important area of research. Modern techniques should focus on the ability to integrate node and edge level characteristics that describe actors and their relationships beyond network structure, and should consider dynamic networks where various aspects of the network can evolve. The spread of information in networks is an area where increased understanding and development of new techniques and algorithms can have implications for a wide range of applications such as improving healthcare and diffusing innovations. In this area, an important task is identifying influential actors who are able to effectively disseminate information, and susceptible actors who are very likely to accept information. Communities within networks are an important driver of information spread and form a basis for developing computationally efficient methods.
This proposal develops community-based methods for predicting influence and susceptibility and identifying influential and susceptible actors. Initial evaluations for these methods which show the utility of focusing on community information are described. The proposed research focuses on three areas of inquiry:
- Understanding how accurate predictions of influence and susceptibility can be made using community-based methods,
- Investigating the correlations between actors who are influential and susceptible, and
- Researching metrics and methods to characterize and compare communities with respect to information spread.
Insights in these three areas will allow for a deeper understanding of how information spreads in networks and can inform strategies for increasing the computational efficiency of a range of network analysis and modelling methodologies.
Questions? Contact Pranav Pamidighantam.