Twitter data for research: from understanding relationships to spotting the Aurora Borealis

Thursday, 30 July 2015

We often say that Twitter data has unlimited value and near limitless application. Twitter data represents the largest archive of human behavior in existence, and what we can learn from it is virtually unlimited. We’ve seen how the Twitter Ecosystem is thriving through the creation of exceptional business products. Another area worth highlighting is the use of Twitter data in the field of research. We continue to be blown away about what people can learn using aggregated public Twitter data.

We’re seeing Twitter data used in a wide range of applications, whether to better understand the concerns of constituents of a city or predict where Northern Lights will appear. To show you exactly what Twitter data is capable of, here’s a sampling of the types of research we’ve seen published so far this year.

Earlier, we posted about Dr. John Brownstein (@johnbrownstein) and his research using Twitter data to better understand public health. He’s far from alone in his research efforts. Another team used machine learning to look at thousands of public Tweets to look for unsuspected drug interactions and unknown side effects beyond data found in clinical trials.

Not only can Twitter data be used for public health; it can be used to gauge citizen satisfaction. The City of Santa Monica launched the Wellbeing Project using public Twitter data (along with that from other social media) to measure the happiness and well-being of its constituents. Using this data, the city was able to help better understand the top concerns of its citizens. The top areas of concern according to examining Tweets were financial, i.e. economic opportunities, jobs, and opportunities. With this data in hand, the city is developing programs to address these concerns

Studying constituents via Twitter data isn’t the only way to better understand the health of a city. Researchers used geotagged Tweets to delve into the travel patterns between Louisville’s West End and the rest of the city. They were hoping to understand racially-based spatial segregation between the two areas, and found that the lines between the two are more fluid than had been assumed.

Since people share major milestones on Twitter, it’s no surprise that Twitter data can even be used to look at how behavior in relationships changes over time. The paper “She Said Yes! — Liminality and Engagement Announcements on Twitter” looked at how after getting engaged, couples’ public Tweets were behavioral and linguistic (they were more apt to talk about “we,” and partake in couple-centric activities).

Twitter data can even indicate where an Aurora Borealis may be spotted helping scientists have more sightings. After an electromagnetic storm in 2011 brought a flurry of Tweets with spottings of the Northern Lights even far down in the south, NASA scientist Elizabeth MacDonald created Aurorasaurus as a way to document sightings and to verify Tweets. A team of researchers led by Nathan Case found that Twitter data not only is a great indicator for where Aurora Borealis will be published but helps identify key characteristics of the light show including color.


We hope this newly launched blog can serve as a way to regularly highlight the types of research that is happening on Twitter data. If you’re publishing some interesting research on Twitter data, let us know via @Gnip.