University of Notre Dame
Browse

Navigating Social Media Narratives

Download (2.8 MB)
thesis
posted on 2023-06-30, 00:00 authored by Nicholas Botzer

As social media continues to shape modern society, understanding how individuals engage in these online spaces and form overarching narratives is crucial. Narratives have a significant impact on people’s lives by influencing public opinion, and shaping society’s values. Although these narratives are often known to occur, it can be challenging to completely understand them and what aspects of messaging resonate with people. With the recent advances in deep learning and natural language processing, methods can now be developed to help understand narratives on social media from a variety of perspectives.

In this dissertation I investigate the development and impact of narratives on social media by examining three key dimensions: moral judgments, conversational flow, and user intent. Moral judgments have become a predominant force for individuals to express their outrage and gather support for social and cultural issues. These judgements are often formed into a narrative to highlight differences between groups but little has been done to understand broad patterns that may occur. Narratives also emerge from the way conversations flow on social media, where like-minded individuals form communities that often develop into echo chambers of discussion These groups discuss topics in a repetitive manner, and I seek to understand this by modeling conversational flow with a graph-based model. Finally, I study narratives from the perspective of user intent. Intent classification is a challenging problem that often requires laborious data annotation to be applicable to targeted domains. Understanding user intent on social media is lacking due to this barrier. To alleviate this issue, I approach the problem in a semi-supervised manner to make it easier to achieve high performance while requiring minimal annotation efforts. This will allow small annotation efforts to be done to conduct studies of user intent on social media for targeted issues. The findings provide valuable insights into the dynamics of online discourse and the factors that drive the formation and propagation of narratives on social media.

History

Date Modified

2023-07-13

Defense Date

2023-05-23

CIP Code

  • 40.0501

Research Director(s)

Timothy Weninger

Committee Members

Meng Jiang Kevin Bowyer Yonatan Bisk

Degree

  • Doctor of Philosophy

Degree Level

  • Doctoral Dissertation

Alternate Identifier

1390203931

OCLC Number

1390203931

Additional Groups

  • Computer Science and Engineering

Program Name

  • Computer Science and Engineering

Usage metrics

    Dissertations

    Categories

    No categories selected

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC