Measuring Responsiveness in the Online Public Sphere for the 2016 U.S. Election: Concepts

TL;DR

This paper proposes a framework using multivariate Hawkes Processes to measure influence among political actors in online discussions during the 2016 U.S. election, aiming to understand narrative formation.

Contribution

It introduces a novel conceptual framework and analytic pipeline for quantifying influence and responsiveness among election players using online data and Hawkes Processes.

Findings

Preliminary influence networks inferred from Twitter and news data.

Demonstrated feasibility of the analytic pipeline for online public sphere analysis.

Identified key influence patterns among election stakeholders.

Abstract

The election narrative is formed under the competitions of ideas among critical players involving politicians, news media, public influentials, and the general public. Untangling the complex process of narrative formation, however, is no easy task due to implicit influences among the key players. This paper outlines a conceptual framework to untangle this complex process. We propose the problem of measuring "responsiveness" that quantifies a player's influence on another given a specific election topic over time. In particular, we make use of multivariate Hawkes Process to infer the influence network between pairs of election players. We demonstrate an early version system of analytic pipeline of online public sphere discussions from data ingestion, influence inference, to visualization. The paper concludes by showcasing some preliminary results based on Twitter and news media election-related contents from July to October 2015 and discussing plans for future research.