What we do better than anyone else…
What we do better than anyone else…
Next up: “Fake News & Other AI Challenges” in Vienna. Texifter Founder & CEO Dr. Stuart W. Shulman will present new applications of the DiscoverText platform to the problem of humans and machines learning.
Please feel free free to share these resources with your colleagues. Permanent links on websites and blogs are incredibly helpful for us, as are reviews.
Support Articles: https://texifter.zendesk.com/
Scholarly Mentions: https://discovertext.com/
Historical Twitter Estimates & Access: https://sifter.texifter.com/
DiscoverText Explained: https://vimeo.com/274453845
DiscoverText Explained (Arabic): https://vimeo.com/278342020
DiscoverText Explained (French): https://vimeo.com/275834878
DiscoverText Explained (Spanish): https://vimeo.com/275595481
Capterra User Reviews: https://www.capterra.com/p/
Note: This is a first in a series of posts about ongoing work initiated by @stuartwshulman for a keynote talk at #screentimebu2018 titled “Fear and Loathing in American Politics: Emotion Detection in Twitter Data”
Part One: Why Fear?
Fear can be a profound political force. After World War II, research suggested fear in America was widespread and broadly centered around Cold War tensions and the existence of atomic weapons. A culture of paranoia tapped into racial, sexual, and other fears of crime and physical harm from “strangers.” The events of September 11, 2001 generated fear arising from international threats, such as terrorist attacks and biological or nuclear weapons.
Today, there are widespread fears related to President Trump, and to major political, social, environmental, legal, and cultural upheavals. The retirement of Justice Kennedy exacerbates prevailing fears among moderates and liberals, while a pending report from Special Counsel Mueller has even the White House staff worried. A recent survey analysis of fear finds that respondents in the United States are more fearful and anxious than ever before .
As a guide to current prevailing fears, we reviewed the recent Chapman University Survey , which indicates political fear in the U.S. has shifted significantly from previous years. Political issues rank among the top fears reported by those surveyed in the United States, including “political unrest and uncertainty in the wake of Donald Trump’s election as president.” The survey found that political fears include involvement in another world war (48.4%) and North Korea using their weapons (47.5%).
With newer communication technologies, social media platforms are common-place, supporting the proliferation of user-generated content. On these platforms, particularly on Twitter, users share stories, discuss contemporary fears, and express concerns. We leverage Twitter data using Texifter ‘s text analytics software, DiscoverText, to augment our survey-based understanding of contemporary American political fear. In the process, we are refining our approach to how we collect Twitter data as well as how humans and machines “learn” together.
George Gallup revolutionized survey research and coined the phrase “pulse of democracy” to describe how polling could inform politicians between elections. We hope to similarly transform Twitter research to produce another facet of political vital signs. While it is tempting to call it a fear index, we are not creating the usual ordinal scale. We are not counting retweets or likes, nor are we measuring purported influence; in fact we intentionally ignore both of these common Twitter metrics. Rather, we see this exercise as generating a map of the ideological highways that American political fear may travel upon.
Our research generally supports the findings of the Chapman Survey. The most frequent fears expressed via Twitter include government corruption, the erosion of democracy, climate change, war, personal safety, and economic fear. A new experimental focus on first-person expressions of fear on Twitter, nevertheless provides a vivid glimpse into the psyche of real-life, with realtime, unfiltered expressions of fear. In part two of the series, we explore what it means to isolate first person expressions using a combination of search terms, human coding, and machine learning.
We just started collecting Gnip Twitter data for the rule “White Helmet” OR whitehelmet recently, accumulating about 38,000 Tweets over a few days.
The most viral RTs are shown here:
We invite you to join an ad hoc group to study the White Helmet campaign. If you would like free software, free training, and a chance to shape future research using advanced tools for human and machine learning, please send me an email today.