Online Program Home
My Program

Abstract Details

Activity Number: 422 - Contributed Poster Presentations: Social Statistics Section
Type: Contributed
Date/Time: Tuesday, July 30, 2019 : 2:00 PM to 3:50 PM
Sponsor: Social Statistics Section
Abstract #307103
Title: Measuring Polarity from News Sources: a Topic Modeling Approach
Author(s): Shane Bookhultz* and Nathan Wycoff
Companies: Virginia Tech and Virginia Tech
Keywords: Topic Modeling; Temporal Latent Dirichlet Allocation; Polarization; Computation
Abstract:

The world is a polarized place due to people that hold firm averse opinions. In recent years, this subject has come to light due to the emergence of social media and divisive journalistic campaigns. To demonstrate the increasing level of polarity, we analyze daily news articles by expanding on a high-dimensional, temporal variant of the Latent Dirichlet Allocation (LDA) algorithm. This algorithm allows us to discern evolving daily topics and the intensity and accelerated rate that the topics enter and leave the media. Furthermore, the temporal LDA algorithm can be computationally expensive, so we implement techniques to improve the efficiency of the algorithm. Since the veracity of news sources are at times questionable, we address polarizing topics to determine the levels of agreement between these sources. To measure the polarization, we propose a new metric for assessing polarity between and within news sources and articles, based on entropy metrics. The performance of our temporal LDA algorithm and polarity metrics are compared with other competing methods. Through this work, we assess the nature and origins of various polarizing sources and evaluate their impact on society.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program