The National Institute of Standards and Technology launched a quarter-million dollar privacy technology competition this month aimed at making it more difficult to trace large data sets back to individual users.
Big data is extremely useful for any number of projects. The problem is that even if you remove the names from large data sets, data can be traced back to individuals. In any zip code, for example, there may be only a few people matching the same age, gender, weight and salary profiles. The more fields you add, the easier it would be to trace.
The NIST competition focuses on public safety information using geolocation data, but the same techniques could transfer to the private sector said competition organizer Gary Howarth.
"We're focusing on temporal and geographic data: a person being tracked over a period of time, like a police officer," he said. "But there are huge applications elsewhere. Think about all the cell phone applications that collect data."
There are both legal and business reasons that private firms would be interested in developments in privacy-protecting technology for big data.
"From a practical standpoint, something Apple and others have shown is that customers will select for privacy," said Jeffrey Vagle, an assistant professor of law focused on privacy, cybersecurity and the ethics of technology at Georgia State University.
"But there are also legal issues in the [the European Union privacy law] GDPR, Brazil and India that go well beyond what companies are used to in the United States," he added.
While there have been several attempts at creating what's known at differential privacy in the past, the technologies sometimes don't hold up as data sets become more specific. The best technologies are not generally available to the public, and there are still unsolved problems in what metrics to use to evaluate how successful algorithms are.
Howarth says the competition would address all three problems with three prize paths: one gauging general success, one in developing metrics to determine success, and one in creating usable open-source repositories of differential privacy technology. Competitors will be able to apply for mentorship to improve the usability of their open-source repositories.
The competition started Oct. 1. It is the second NIST differential privacy contest, following one in 2018.