Date of Award

5-7-2020

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Information Science

First Advisor

Nitin Agarwal

Abstract

Social media is widely used to express views and share opinions with others. With the availability of inexpensive and ubiquitous mass communication tools like social media, creating narratives, false information and propaganda is both convenient and effective. Social media users leverage this platform to further their views by framing narratives and participating in online discourse. Almost all events, issues, crises are discussed on social media. Blogs are a good source of data for sociologists and political scientists to gain situational awareness by tracking different opinions, political views, and narratives being shaped. Blogs, unlike other social media platforms, are not regulated by any authority and have no restriction on character limit, provide bloggers with not only space for richer content but also serve as a platform for agenda-setting and content framing abetting development of narratives. Fundamentally, narratives are described with actors and their actions in a given discourse. We propose a novel framework that computationally identifies actors/actions with the help of NLP techniques that include - POS tagging, chunking to obtain noun/verb phrases, and define grammar rules that captures phrases of noun/verb, and finally filter phrases/sentences based on TF-IDF to generate narratives. To evaluate the efficacy of the proposed model, we validate the narratives generated with the human annotated narratives. The results for the representative sample of the data indicate the accuracy of our framework as 66.8%. Further, we apply the same model to topic streams and identify narratives within each topic. For a given blog post, our model generates dominant narrative first and then less dominant ones in the decreasing order of their rank. Our proposed framework can help social scientists identify narratives computationally eliminating human effort reasonably. This enables the analysts to learn what resonates with the community and if those interests and views are changing with time under the influence of exogenous factors or events. The results from this study could also be used to build counter narrative measures against propaganda campaigns.

Download

Included in

Social Media Commons

Share

COinS