A relatively complete picture of our movements can be reconstructed from anonymized data generated by mobile phones by analyzing the movements of our social contacts, researchers say.
These anonymized details can also be used to reveal the nature of relationships between people, such as whether they are casual acquaintances, co-workers, or friends or family, scientists add.
Personal data often gets anonymized by stripping it of identifying details such as names and home addresses. This "metadata" often get shared, and underlie popular services such as Google's real-time traffic monitoring, which shows road conditions in more than 50 different countries.
However, anonymized data can still give away a great deal about individuals. For example, scientists recently found that anonymized cell phone data could be better at identifying users than fingerprints, and that anonymized credit card data could easily be used to identify its users.
Now researchers find that anonymized phone call records can reveal a great deal about a person by looking at when and where they may interact with others.
The scientists analyzed 1 billion records logged whenever mobile phone users in three cities in two different industrialized nations connected to a cell phone tower, such as when they placed a call or sent a text. Each record included the time of the event, a unique anonymous ID for the call-maker and call-recipient, and the identity of the tower used by the call-maker and sometimes the call-recipient, which gives a rough estimate of where one or both are located.
The investigators focused on visitation patterns—what places individuals visited, and when they visited them. They found that an individual's visitation patterns were far more similar to those of social contacts—of people an individual called who called that individual back—than to random strangers. As such, they could study the movements of a person's social contacts "to reconstruct a relatively complete picture of that person's movements," says lead study author Jameson Toole, a data scientist at MIT. "About 20 percent of the trips people made appear motivated by social contacts."
In addition, by looking at how similar visitation patterns were between individuals and their social contacts, three kinds of social contacts emerged. One group is likely made up of friends or family, with similar patterns in the evening and on the weekends. Another group is similar during working hours but neither evenings nor weekends, and are probably co-workers. The last group had uniformly low levels of similarity, and were dubbed acquaintances.
One implication of this work for privacy "is that no matter how off the grid I can be, if someone has information on my social contacts, they can infer a bit about me," Toole says. "There doesn't seem to be a good solution to this matter of privacy—solving it would mean not being friends with anyone you have anything in common with."
Toole suggests that the research could lead to a better understanding of where people want to go, and that could help improve urban transportation. "A good model of where everyone wants to go could help forecast travel demands to help set bus routes, for instance," he says. "Also, new services such as Uber and Lyft and Zipcar that involve ride-sharing and car-sharing are predicated on matching people to others they trust and going places they all want, and matching people who have similar mobility behavior could help with those."
Toole and his colleagues detailed their findings 25 February in the Journal of the Royal Society Interface.
Charles Q. Choi is a science reporter who contributes regularly to IEEE Spectrum. He has written for Scientific American, The New York Times, Wired, and Science, among others.