The Large Apple versus The Large Orange. The Metropolis of Goals versus The Metropolis of Angels. I’m referring, after all, to the continuing rivalry between New York Metropolis and Los Angeles. Hilarious “survey” movies and discuss exhibits offers you one image of the cities. My colleagues and I made a decision to take a extra critical have a look at the variations between the cities, so we studied what everybody else was speaking about – on Twitter.
We got down to reply a easy analysis query: Are people who find themselves positioned close to one another more likely to tweet about related issues? To take action, we analyzed hundreds of thousands of GPS-enabled tweets throughout New York Metropolis and LA. Such a research – large quantities of social media visitors by location – is beneficial for greater than monitoring popular culture memes in numerous cities. It may very well be useful for understanding many facets of city life, together with the results of the COVID-19 pandemic.
If we have been contemplating the case of a single, small group that takes delight in native occasions, celebrities and tradition, the reply to our analysis query can be a convincing “sure.” One problem in evaluating two giant, worldwide cities is the truth that globalization has led to unprecedented interplay amongst a number of cultures and peoples, together with Starbucks and McDonald’s seemingly in each metropolis on the planet.
For cities which might be worldwide but additionally take delight of their uniqueness, the secret is teasing out the extent to which native qualities or world tradition dominate tweeting conduct. We designed our strategies to be exact sufficient to account for the truth that, opposite to the enjoyable movies, New York Metropolis and LA are fairly related. Each have excessive housing prices, well-known academic establishments, hospitals, museums and different cultural institutions, and residents who are likely to vote Democratic.
Outline ‘shut’ and ‘identical’
Our research tackled two issues: There’s no easy definition of “shut collectively,” and it’s tough to say whether or not two tweets are about the identical matter. We mixed a number of definitions of “shut collectively,” starting from individuals positioned in the identical metropolis to the space in miles between their coordinates, utilizing a typical system from spatial sciences.
Minda Hu, CC BY-ND
It’s tougher to find out whether or not two tweets are speaking about related issues. Searching for frequent hashtags would possibly suffice, however sadly many individuals don’t use hashtags or use totally different hashtags when speaking about the identical factor. To beat this downside, we used state-of-the-art pure language processing know-how. Algorithms developed on this area learn and interpret sentences in a fashion just like the way in which people do, and they can take care of nuance.
We used this know-how to group tweets into clusters of subjects. We then studied whether or not tweets falling inside the identical cluster have been additionally from individuals who have been shut collectively primarily based on their GPS-enabled tweets. This allowed us to find out, for instance, that clusters containing art-related phrases and phrases tended to come up extra usually in New York than LA.
Well being and wealth versus artwork and representing
Even earlier than we checked out who tweets about what, we discovered tweeting throughout New York Metropolis to be extra evenly unfold, whereas in LA, extra tweeting occurs in wealthier areas, together with Calabasas – dwelling to Kim Kardashian – Palos Verdes, West Hollywood and the coastal areas.
We additionally discovered that New Yorkers referred to themselves and their metropolis way more usually than Angelenos did. On a per capita foundation, New Yorkers like to speak about artwork, whereas Angelenos like to speak about well being care and hospitality.
LA generates extra tweets than New York all through the day, regardless of having a smaller inhabitants, however from 8 p.m. to five a.m. native time, the 2 have comparable numbers of tweets. Tweeting in New York Metropolis rises sharply from 8 p.m. to a peak at 9 p.m., whereas tweeting in LA rises steadily from 2 p.m. to a peak at 7 p.m.
Computational social science
Our strategies are a case research within the rising area of computational social science, which goals to seek out insights in distinctive, usually giant, information units utilizing synthetic intelligence fashions and algorithms. In distinction, conventional social science tends to depend on surveys and polls to quantify public notion about a difficulty. Although surveys have some distinct statistical benefits, they are often costly and time-consuming to make use of for amassing high quality information with good response charges.
For instance, Gallup releases new survey information each few months and presently prices US$30,000 for educational licenses. A long time in the past, researchers discovered that financial incentives enhance response charges considerably. Even at the moment, on-line surveys are sometimes accompanied by lottery-based guarantees of receiving an Amazon reward card. Researchers are engaged on combining the advantages of conventional and computational social science.
Zooming into our information, we uncovered some fascinating traits that we hope future analysis will discover. We discovered, for instance, that on a per capita foundation, as crime will increase, so do tweets, not less than on the stage of ZIP codes. Why do high-crime areas tweet extra? We don’t know but, however the development is constant throughout each New York Metropolis and LA.
Tweeting, place and COVID-19
Learning tweeting conduct by location is also helpful for understanding disparate outcomes of large-scale occasions. For instance, our twitter evaluation might assist make clear how the COVID-19 pandemic has affected individuals in other places.
New York Metropolis was hit onerous by COVID-19 early on, exhibiting that even main cities have been affected in numerous methods by this horrible pandemic. New reporting is now exhibiting that even inside cities, socioeconomically deprived communities have been disproportionately burdened.
Lately, we launched a Twitter information set masking 10 of the biggest metropolitan areas in the US to additional research such disparities utilizing computational social science. We’re already utilizing our strategies throughout all of those cities to higher perceive how COVID-19 has affected sure teams, and the degrees of expressed vaccine hesitancy amongst these teams.
Finally, we hope to make use of our strategies with a big set of worldwide metropolises to review city conduct.
[Get our best science, health and technology stories. Sign up for The Conversation’s science newsletter.]