Modern Approaches for the Analysis of Social Media Data

North Carolina State University, Raleigh NC

Investigators

Abstract

This research project will develop statistical methods for the analysis of high-resolution data arising from social media applications. Technological growth has made possible the accumulation of data from social media platforms at unprecedented speed and volume. However, current methods for analyzing these data either lack interpretability, are computationally intense, or require a rigid data regimen. This project will use a flexible modeling framework to extract relevant information on a user's behavior. Although the project primarily will focus on Twitter data, the methods to be developed will be applicable to other social media data, such as Facebook, Instagram, Reddit, or TikTok, or any form of digital interaction. The results of this research will help managers, policymakers, and stakeholders better understand the types of actors they are interacting with on social media. The methods and code created by this project will be made publicly available. The investigators will mentor undergraduate and graduate students and interact with K12 students who are interested in using valid statistical approaches to solve problems arising in online social media. This research project will develop statistical methods for binary and categorical functional data structures. The standard analysis of functional data relies fundamentally on the assumption that the intrinsic functions are continuous over the compact interval. The complexity of social media data, however, requires different assumptions. A Twitter user's posting pattern can be defined as a time series of some feature of the posting activity. The user's data then can be viewed as a binary-valued or categorical-valued random function defined over a time domain and observed at a fine grid point. The project will develop classification methods for non-continuous functional data, propose computationally efficient estimation algorithms, and study their theoretical properties. The methods to be developed will be extended to adapt to multiple binary-valued or categorical-valued curves per subject, acquired in a longitudinal design, as well as to account for additional covariate information. Regression models with categorical-valued functional covariates and their associated significance tests also will be developed. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.