Providing secure access to sensitive data
1 / 5

Providing Secure Access to Sensitive Data - PowerPoint PPT Presentation

  • Uploaded on

Providing Secure Access to Sensitive Data. John M. Abowd Cornell University IASSIST 2010 June 4, 2010. Everybody's Got the Same Problem. Substantial Data Asset Physical Security Layer Access Protocol Public View on Private Data. Firewall. Firewall. Google Search Logs.

I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
Download Presentation

PowerPoint Slideshow about 'Providing Secure Access to Sensitive Data' - oberon

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.

- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
Providing secure access to sensitive data

Providing Secure Access to Sensitive Data

John M. Abowd

Cornell UniversityIASSIST 2010June 4, 2010

Everybody s got the same problem
Everybody's Got the Same Problem

Substantial Data Asset

Physical Security LayerAccess Protocol

Public View on Private Data



Google Search Logs

American Community Survey Micro-data

Research Paper from a Census RDC

Leading Indicators Based on Search Frequencies

But they don t use the same strategies
But They Don’t Use the Same Strategies

  • Microsoft, Google, Yahoo!, Amazon tend to rely on formal privacy methods

  • Scientific validity and future data integrity are corporate assets

  • Most strategies explicitly integrate content and privacy functions

  • Distinguish between types of data releases:

    • Official statistics

    • Special tabulations

    • Public-use files

    • Protocol data access (RDC, AFF Advanced Query)

  • Most strategies isolate the content and disclosure limitation functions

And they don t make the same mistakes
And They Don’t Make the Same Mistakes

  • AOL and Netflix search log releases were fully preventable

  • Information loss from randomized sanitizers is fully quantifiable but scientifically difficult to digest

  • Not as concerned about the public good aspect of data release

  • Research papers viewed as usually safe when based on statistical modeling

  • Concerned about the public good from statistical publications but limited by ad hoc disclosure avoidance procedures that provide security through obscurity

What to do next
What To Do Next

  • Recognize that every public view from official statistics to model output in scientific papers can be assessed for both statistical validity and privacy/confidentiality leakage

    • There is a continuum from official statistics through “need-to-know” privileged access that all research paradigms touch—think more about the substance and less about the form

  • Integrate research teams primarily interested in content with teams primarily interested in privacy protection

    • Be certain that both groups know what the other is doing