13 August 2021

Tapping Twitter’s Potential

Welcome back. In case you were wondering, I don’t have a Twitter account. Yes, I sometimes check friends’ tweets about technical news, and yes, I certainly considered joining the Twitterverse. I even shared my pros and cons in a long-ago blog post, Tweet?.

More important, I want to tell you about a powerful, new tool for curating and analyzing tweets, specifically, about 10% of all the tweets made every day around the globe. I think you’ll find it of interest.

Twitter’s Potential
Paul Lewis, with The Guardian, described Twitter as the digital footprint of things that are happening around the world. And there’s no question that Twitter, founded in 2006, has tremendous promise if one can get by the lies and rumors.

As early as 2009, Jake Coyle, with the Associate Press, wrote: Twitter has in many ways been a boon to the media…Most outlets now have a presence on Twitter with a feed directing readers to their respective sites… But truthfulness remains the biggest problem…False rumors spread daily on Twitter.

An exemplar topic addressed in different research investigations in Twitter’s early years was its use for detecting disease outbreaks. Through the years, Twitter’s value for reporting outbreaks and disasters has been demonstrated repeatedly, though rumors have often interfered (see my post Tweets Misreport Disasters).

All of which brings me to Storywrangler, the tool developed by researchers with the University of Vermont, Charles River Analytics and MassMutual Data Science.

Storywrangler website header (storywrangling.org/).
Storywrangler provides a natural language processing framework that extracts, ranks and organizes tweets, generating frequencies for words (1-, 2- and 3-word phrases), hashtags, handles, numerals, symbols and emoji.

Capturing time-stamped messages and storylines in more than 150 languages, from Twitter’s inception to the present, Storywrangler makes the datasets available through an interactive time-series viewer and as downloadable time-series and daily distributions.

While Storywrangler could, in principle, be modified to operate on other social media platforms, Twitter offers some advantages. One noteworthy example is Twitter’s social amplification mechanisms--retweets and quote tweets, which enable the explicit encoding of popularity.

Visual Comparison of Phrase Popularity

In their published paper, the researchers presented a sample of Storywrangler’s online viewer visualizations. On one graph, they illustrated how different Twitter features capture three global events from the first half of 2020.

Storywrangler visualizations of three global events from first half of 2020: the death of Iranian general Qasem Soleimani; the beginning of the COVID-19 pandemic (term “coronavirus” as yellow, virus emoji, green); and the Black Lives Matter protests following the murder of George Floyd (from advances.sciencemag.org/content/7/29/eabe6534/tab-article-info).
Storywrangler showed a spike of tweets and retweets of the word "Soleimani" on 3 January, when the U.S. killed the Iranian general Qasem Soleimani with a drone strike.

There was a rise of the word "coronavirus" as well as the virus emoji as COVID-19 spread with the beginning of the pandemic in the spring.

And the hashtag "#BlackLivesMatter" jumped on and after 25 May, the day George Floyd was murdered.

Wrap Up

Storywrangler can be used to follow, trace and explore the serious and the not-so serious--disasters, political matters, vaccines, federal reserve actions, entertainment, fashion, dark matter, the dark web--and monitor discourse by or about the famous and the not-so famous.

But as the Storywrangler website warns, Twitter provides a non-representative subsample of utterances made by a non-representative subsample of Earth’s population; and Storywrangler reflects only a random 10% of those messages.

Though Storywrangler doesn’t resolve the lies and rumors that come with Twitter, the tool provides the opportunity to dig a little deeper. That’s at least a head start. Thanks for stopping by.


Twitter’s potential:
- Jake Coyle: abcnews.go.com/Technology/story?id=7979891&page=1
- Paul Lewis: theconversation.com/how-twitter-has-helped-the-emergence-of-a-new-journalism-19841
Study on development of Storywrangler in Science Advances journal: advances.sciencemag.org/content/7/29/eabe6534/tab-article-info
Article on study on EurekAlert! website: www.eurekalert.org/news-releases/527597
Storywrangler website: storywrangling.org/


No comments:

Post a Comment