National ICT Australia (NICTA) is promising big data analytics while preserving privacy through the use of homomorphic encryption.
“Everyone wants to do big data, but users want to maintain their privacy,” said Terry Percival, director of broadband and the digital economy at NICTA, at CommunicAsia2015. “That’s the problem we’ve solved. How can you gain insight from data without actually giving up privacy - use magic? That’s what we call our project: Magic.”
Magic is a tool that measures the trade-off between data fidelity and privacy. It can group together customers and add noise. The larger the group, the less fidelity and the more privacy individual users have. The tool calculates and reports on fidelity of the trade-off through the analytics process.
The other half of Magic is moveable analytics. Algorithms send this grouped, noisy data for analysis to any number of different third parties.
Multiple obfuscated data-sets can be run for big data analytics by different processors, with nobody having access to the full set of data - yet with an outcome useful for detecting outlying data points and anomalies.
NICTA’s first POC is using Magic between a telco and a bank to test how banking information and mobile phone billing history can be combined to give meaningful analytics without either side revealing their customer database to the other. Another POC combined a travel company with a bank to detect suspicious activity.
NICTA is also showing off its business process compliance tool: a new language that formalizes legal compliance checks, rules, and regulations into a new computer language.
Defeasible detonic logic rules could include check once, check persistently, completed once, completed before and so on. Compliance rules such as Sarbanes-Oxley are translated into this language, as are business processes, and the tool can automatically run to see if they comply (and if not, what part of the business process does not). It can also check existing processes with new or proposed laws.