MOSCOW, November 22 (Sputnik) — Researchers at Cornell University have shown that analyzing the tweeting patterns of social network users reveals the unemployment rate of the region in which they live.
“We demonstrate that behavioral features related to unemployment can be recovered from the digital exhaust left by the microblogging network Twitter,” say the scientists in their paper, named Social Media Fingerprints of Unemployment and published by Cornell this month.
The scientists analyzed almost 146 million geo-located messages in Spain, and matched the behavioral patterns exhibited in the Tweets with the unemployment rate of 340 Spanish municipalities over a seven-month period at the height of the Spanish financial crisis of 2012-13. “We consider unemployment to be the most important signal for the socioeconomic status of a region,” they wrote.
The researchers found that several indicators can be used to infer unemployment levels, using data from Twitter. For example a peak of tweeting at the beginning of the working day was exhibited in areas of low unemployment, and a high rate of misspelling corresponded to high unemployment. Positive correlation was also found between unemployment and Tweets containing the words job or unemployment, while the words employment and economy negatively correlated.
In contrast with recent research, which has shown that the level of use in one country compared to another positively correlates with a higher GDP, the study found that the level of Twitter use within a country’s area is indicative of higher unemployment there.
The researchers also highlighted the cost advantage of using such publicly-available data, particularly as economic recession limits the amount countries are willing to spend on data collection. “Most importantly, the immediacy of social media may also allow governments to better measure and understand the effect of policies, social changes, [as well as] natural or man-made disasters in the economical status of cities in almost real-time,” wrote the scientists.
The study is not the first to use Twitter as a basis for social data mining. Earlier this year, a group of researchers from Penn State claimed to have developed “a system for making an accurate influenza diagnosis based on an individual’s publicly available Twitter data,” with 99% accuracy.
