Building an Anonymization Pipeline. Creating Safe Data (ebook)

Building an Anonymization Pipeline. Creating Safe Data (ebook) - Opis

How can you use data in a way that protects individual privacy but still provides useful and meaningful analytics? With this practical book, data architects and engineers will learn how to establish and integrate secure, repeatable anonymization processes into their data flows and analytics in a sustainable manner.Luk Arbuckle and Khaled El Emam from Privacy Analytics explore end-to-end solutions for anonymizing device and IoT data, based on collection models and use cases that address real business needs. These examples come from some of the most demanding data environments, such as healthcare, using approaches that have withstood the test of time.Create anonymization solutions diverse enough to cover a spectrum of use casesMatch your solutions to the data you use, the people you share it with, and your analysis goalsBuild anonymization pipelines around various data collection models to cover different business needsGenerate an anonymized version of original data or use an analytics platform to generate anonymized outputsExamine the ethical issues around the use of anonymized data Spis treści:
Preface
Why We Wrote This Book
Who This Book Was Written For
How This Book Is Organized
Conventions Used in This Book
OReilly Online Learning
How to Contact Us
Acknowledgments
1. Introduction
Identifiability
Getting to Terms
Laws and Regulations
States of Data
Anonymization as Data Protection
Approval or Consent
Purpose Specification
Re-identification Attacks
AOL search queries
Netflix Prize
State (...) więcej Inpatient Database
Lessons learned
Anonymization in Practice
Final Thoughts
2. Identifiability Spectrum
Legal Landscape
Disclosure Risk
Types of Disclosure
Learning something new
Dimensions of Data Privacy
Linkability
Addressability
Identifiability
Inference
Re-identification Science
Defined Population
Direction of Matching
Sample to population (public)
Population to sample (acquaintance)
Structure of Data
Cross-sectional data
Time-series data
Longitudinal or panel data
Multilevel or hierarchical data
Overall Identifiability
Final Thoughts
3. A Practical Risk-Management Framework
Five Safes of Anonymization
Safe Projects
Primary and secondary purposes
When to anonymize
Safe People
Recipient trust
Acquaintances
Safe Settings
Risk matrix
Safe Data
Quantifying identifiability
Safe Outputs
Invasion of privacy
Five Safes in Practice
Final Thoughts
4. Identified Data
Requirements Gathering
Use Cases
Data Flows
Data and Data Subjects
Data subjects
Structure and properties of the data
Categories of information
From Primary to Secondary Use
Dealing with Direct Identifiers
Realistic direct identifiers
Dealing with Indirect Identifiers
From Identified to Anonymized
Data (anonymization) processors
Controlled re-identification
Mixing Identified with Anonymized
Functionally anonymized
Five Safes as an information barrier
Applying Anonymized to Identified
Final Thoughts
5. Pseudonymized Data
Data Protection and Legal Authority
Pseudonymized Services
Legal Authority
Legitimate Interests
A First Step to Anonymization
Revisiting Primary to Secondary Use
Analytics Platforms
Remote analysis
Secure computation
Synthetic Data
Differential privacy
Biometric Identifiers
Secure computation of genomic data
Final Thoughts
6. Anonymized Data
Identifiability Spectrum Revisited
Making the Connection
Anonymized at Source
Additional Sources of Data
Pooling Anonymized Data
Pros/Cons of Collecting at Source
Methods of Collecting at Source
Safe Pooling
Access to the Stored Data
Feeding Source Anonymization
Final Thoughts
7. Safe Use
Foundations of Trust
Trust in Algorithms
Techniques of AIML
Classical machine learning
Neural networks
Technical Challenges
Algorithms Failing on Trust
Rogue chatbot
Predicting criminality
Principles of Responsible AIML
Governance and Oversight
Privacy Ethics
Data Monitoring
Final Thoughts
Index mniej