A team from the Pacific Northwest National Laboratory and the University of Washington have collated the results of hundreds of separate studies and found that the things we post on social media can be used to predict the future.
Analysis of social media, including Facebook, Twitter, and Instagram allows experts to predict civil unrest, disease outbreaks, and even the weather.
The findings were published in a paper on Arxiv. It states that social media can be used in a number of ways to predict future events, and much of it requires only being observant to the buzz and chatter taking place on such sites.
“If SM users are reacting to and talking about events in real time, one might imagine that users are also talking about and reacting to events that they anticipate will happen in the future,” the report says.
“This raises the interesting possibility that SM data might be useful for forecasting events: making predictions about events that have yet to occur.”
The 55-page research on SM forecasting spans a wide range of topics, including outcomes of elections, fluctuations in the stock market, disease outbreaks and possible detection of other “real-world” threat, such as “natural disasters, security events, and political uprisings.”
Civil unrest in Brazil was predicted with 80 percent accuracy. In Venezuela, the model did not work so well, it predicted civil unrest only 50 percent of the time.
The model was able to predict when diseases would spread, such as norovirus. It was also used to predict the weather.
Of 25 types of crime examined by the researchers, 19 were somewhat predictable.
Mental health diagnoses could also be performed. Researchers found that depression was accurately diagnosed, also with 80 percent accuracy.
Researchers learned that certain hashtags were used by people. The frequency of their use and the intensity of the traffic also provided critical clues.
However, not every topic can be foreseen by simply picking up on social media behavior. The model was useless in predicting sports scores and the stock market.
The findings would be an interesting footnote if not for the wide range of abuses that could result. Corporate interests are already seeking the technology for their own use. If businesses are better able to predict consumer behavior, they can position themselves to profit, or at least minimize losses.
Researchers think they can improve the model with more data and more time.