A1743
Title: Media bias and polarization using a Markov-switching latent space network model
Authors: Antonio Peruzzi - Ca' Foscari University of Venice (Italy) [presenting]
Abstract: The news consumption landscape has drastically changed in the last decades. Web 2.0 and social media are re-shaping the way in which news pieces are consumed and produced. Some old questions renew in such a scenario. One of these is whether and to which extend news outlets bias information. We propose a new dynamic latent-space model (LS) for news outlets in which we exploit both time-varying online duplication-network data as well as textual contents from published articles to measure media bias over time. Within our model, the latent-space positions of news outlets have a proper interpretation respectively in terms of media slant and online engagement. The aim is twofold: making advancements both concerning the analysis of the timely evolution of audience duplication networks and concerning the determination of media slant and polarization by exploiting both textual and network-structure information. The developed model is applied to a Facebook dataset consisting of the information provided by Italian news outlets in the years 2015 and 2016. Eventually, the latent positions of the news outlets over time is analyzed and discussed.
