Researchers are turning to Twitter to track immigrant migration

More than 250 million people migrated away from their birth country in 2017, according to the United Nations. However, tracking migration through surveys, like an official census, is costly and can take years to complete. To answer those concerns, researchers from the Institute for Cross-Disciplinary Physics and Complex Systems have developed a new method—tracking migration using data from the social media platform Twitter, which provides more frequent, nuanced, and perhaps more accurate information.

The researchers analyzed thousands of tweets to develop a model to predict whether individuals are likely natives or immigrants to the places they live, pulling information about users and their posts to make a determination. For example, if a user regularly tweets in German from an address in Philadelphia, the model would label the user as a German immigrant.

Researchers aggregated that data to identify 35 immigrant communities in 50 cities around the world. They then predicted, through user actions, the neighborhoods in those cities where each likely immigrant community lives. If a large portion of the city’s non-immigrant residents also lives in those neighborhoods, the immigrant community is considered to be well-integrated into the city.

Through the research, London rose to the top and took the title of the most integrated city. “We found that in the central part of the city [Chelsea, Kensington, and Marylebone], communities such as Russian, French, and Arabic spread almost uniformly around the area,” says Fabio Lamanna, a civil transportation engineer who led the research team.

“On the other side, there are still some spatial segregation phenomena in the Greater London Area, where, for example, Turkish people are, [on] average, uniformly spread, but with a peak of density in the North-East area of the British capital.”

The researchers also found patterns and distinctions among immigrant groups as a whole, regardless of the cities where they settled. Korean, West-Slavic, and Dutch communities tend to receive the highest integration scores. Spanish-speaking immigrants make up the most common immigrant communities in world cities and generally have high integration scores.

Relying on Twitter data does have drawbacks. Researchers believe communities with fewer third- and fourth-generation immigrants, who tend to be older, aren't as represented on social media. Twitter is banned in China, so no Chinese cities appear in the results. Weibo has instead become a popular alternative, both in the country and with Chinese immigrant communities. For example, New York, which has an extensive Chinese community, shows no Chinese users in the research, and cities without Twitter data, such as Beijing, were excluded from the results.

Still, by using Twitter data, researchers believe they will be able to analyze immigrant migration more regularly and cost-effectively—thereby providing up-to-date insights to support informed policies and discussions on migration around the globe.