New version of Aircloak Insights now available

We are proud to announce the immediate availability of our latest release of Aircloak Insights. This release features improvements in:

• Performance: Reductions in query runtime and memory consumption of up to 60 and 75% respectively
• NoSQL Support: Ability to perform nuanced offloading of computations and JOINs to the database server when recent versions of MongoDB are used
• BI Tools Support: Initial support for querying Aircloak Insights with Tableau, the leading tool used by Analysts
• Core Algorithms: Improvements to the underlying anonymization technology allows us to produce higher quality insights without compromising on the strength of anonymization

The improvements to the core algorithms are what we want to focus on in this blog post.

Aircloak Core Algorithms Primer

Anonymization is a process whereby a dataset is transformed to the point where no data can be tied back to an individual. In traditional static anonymization this transformation can consist of outright removing parts of the data and making other parts coarse-grained. Query-by-query approaches like Diffix (used by Aircloak Insights) and Differential Privacy add statistical noise to individual query results instead.

The main differentiator between the approaches is the amount to which utility is reduced. Quality of insights and bullet proof anonymization were our main goals when developing Diffix. These goals are achieved by deeply understanding the semantics of the queries being run, allowing us to tailor the anonymization to the query as well as the data being queried.

Changes in this release

The noise Aircloak Insights adds to query results has always been low, but in this latest release I am proud to announce that we have managed to reduce it by further 20% for representative queries. This milestone achievement was reached at no cost to the strength of anonymization. Analysts can now surface information from structured and unstructured data, obtaining analytic insights that would get lost when using alternative technologies.

This release highlights one of the benefits of query-by-query anonymization systems over traditional anonymization. A traditionally anonymized dataset will never be safer than it was on day one, nor will it yield richer insights than it could at the time of the initial anonymization. In contrast, Aircloak Insights allows you to improve query and anonymization performance with each new release!

If you have questions about Aircloak Insights or suggestions for future releases, contact us at @aircloak. If you are interested in a demo or have general questions, please email us at solutions@aircloak.com.