The changelog for current versions of ARX is available on our GitHub page.

ARX 3.7.1


  • Prevent exception when using Differential Privacy

ARX 3.7.0

New Features

  • Core: Added method for clustering-based microaggregation
  • Core: Added method specifying for functional transformation rules for dates
  • Core: Added methods for risk assessment using wildcard matching
  • Core: Added methods for applying local data transformation schemes
  • Core: Added new types of privacy-preserving classification models
  • Core: Added new methods for analyzing the accuracy of privacy-preserving classification models
  • Core: New method for optimizing output data to improve the accuracy of privacy-preserving classification models
  • Core: New methods supporting the quantitative analysis of the quality of output data
  • GUI: Added wizard for creating transformation rules for dates
  • GUI: Added dialog for local data transformation, cell suppression and microaggregation
  • GUI: Added new views for visualizing accuracy estimates of classification models
  • GUI: New view for quantitative analyses of data quality


  • Core: Improved performance of t-closeness for ordered attributes
  • Core: Reduced memory consumption
  • GUI: Improved control over asynchronous analyses
  • GUI: Improved execution time estimates for long-running tasks
  • Various minor improvements


  • Various minor bugfixes

ARX 3.6.0

New features

  • Core: Implement hierarchy builder for dates
  • Core: Implement new privacy models (basic and enhanced beta-likeness)
  • Core: Implement method for generating reports
  • GUI: Implement dialog for building hierarchies for dates


  • GUI: Add locale to data types
  • GUI: Improve performance of preview
  • GUI: Show attribute types when displaying features and classes
  • Core: Implement additional method for optimizing output data
  • Core: Extend summary statistics
  • Core: Add f-score to classification analysis
  • Various minor improvements


  • Core: Fix domain share for interval-based transformations
  • GUI: Fix memory and resource leaks
  • GUI: Fix row height of table
  • Various minor bugfixes

ARX 3.5.1


  • GUI: Fix bug which prevented the use of local recoding
  • GUI: Fix bug which prevented the use of delta-presence
  • Core: Handle floating point issues in game theoretic cost/benefit model

ARX 3.5.0

New features

  • Core: Implement new method for considering the impact of microaggregation on data quality
  • Core: Implement game-theoretic approach for optimizing the profitability of publishing de-identified data based on cost/benefit analyses
  • Core: Implement record-level entropy-based model for measuring information loss
  • Core: Implement corrected Grassberger estimator for entropy-l-diversity
  • Core: Implement t-closeness for numeric attributes
  • Core: Implement alpha-distinction and beta-separation for detecting quasi-identifiers


  • GUI: Improve synchronization of tables
  • Various minor improvements


  • GUI: Make calculations of risks deterministic to prevent inconsistencies
  • GUI: Fix import from relational databases
  • Core: Fix import from relational databases
  • Various minor bugfixes

ARX 3.4.2


  • Fix JDBC import dialog
  • Make sure that all data is trimmed during import
  • Fix corruption of files when creating projects with empty hierarchies

ARX 3.4.1


  • GUI: Fix CSV and Excel import

ARX 3.4.0

New features

  • Core: Implement sane method for analyzing classification accuracy
  • Core: Implement methods for analyzing the distribution of re-identification risks
  • Core: Implement new privacy model: estimated k-map
  • Core: Implement generalization/suppression factor for AECS quality model
  • Core: Implement generalization/suppression factor for Non-Uniform Entropy quality model
  • GUI: Implement view for analyzing precision and recall of a statistical classifier
  • GUI: Implement method for top/bottom coding
  • GUI: Implement views for analyzing the distribution of re-identification risks


  • Core: Make suppression string a constant
  • Core: Improve implementation of entropy-l-diversity
  • GUI: Improve views for analyzing classification accuracy
  • GUI: Show whether the currect solution is a global optimum
  • GUI: Automatically add transformations of interest to the clipboard
  • GUI: Use MySQL connector licensed with Oracle FOSS license
  • Tests: Redesign path structure for loading data
  • Tests: Add test cases for anonymization with data subsets
  • Tests: Add test cases for risk-based anonymization
  • Tests: Add test cases for local recoding


  • Core: Do not use suppressed records when estimating risks
  • Core: Fix resource leak when loading data
  • GUI: Handle differential privacy correctly
  • GUI: Fix error when reading/writing projects with empty hierarchies
  • GUI: Correctly show tooltips
  • GUI: Allow using quasi-identifiers with no hierarchy assigned
  • GUI: Fix errors with backwards compatibility
  • Tests: Make differential privacy deterministic

ARX 3.3.1

New features

  • Implemented derived risk metrics, including measures for prosecutor, journalist and marketer risk.
  • Implemented new perspective in the GUI for analyzing these risks.
  • Implemented methods for analyzing data mining utility with logistic regression.
  • Implemented new perspective in the GUI for analyzing classification accuracy.
  • Implemented new privacy model k-map.
  • Implemented new privacy model d-disclosure privacy.


  • Improved visualization of percentages in the GUI.
  • Improved export of data from tables in CSV format.
  • Improved utilization of the clipboard in the GUI.
  • Updates some labels and messages.
  • Improved help with new content.
  • Improved detection of HIPAA identifiers. Display confidences.
  • Improved editing of population parameters.
  • Improved configuration of parameters of privacy models.
  • Improved multi-threading in the utility analysis perspective.
  • Improved process of solving equation systems for the Pitman model.
  • Improved rendering of numbers in the whole GUI.
  • Improved performance of hashing. Partial unrolling.
  • Various minor improvements.


  • Fixed utility measurement when using differential privacy.
  • Fixed file-handling, check if files exist.
  • Fixed a bug with backwards compatibility.
  • Fixed calculation of geometric mean in summary statistics.
  • Fixed calculation of geometric mean in micro-aggregation.
  • Various minor fixes.

ARX 3.3.0

  • Added method for local recoding with k-anonymity, l-diversity and t-closeness
  • Added example project file with high-dimensional data
  • Added methods for detecting HIPAA identifiers
  • Added implementation of (e,d)-differential privacy
  • Added utility measures KL divergence and Ambiguity
  • Improved performance of maintaining the solution space
  • Implemented critical bugfix to ensure correct classification of the solution space
  • Added various minor bugfixes and improvements

ARX 3.2.1

  • Fix risk-based anonymization and analysis
  • Fix normalized entropy measure

ARX 3.2.0

  • Added support for anonymizing high-dimensional datasets
  • Changed GUI to support high-dimensional datasets
  • Added new utility measures: normalized non-uniform entropy and ambiguity
  • Implemented implicit representation of the solution space: ~10% performance improvement and significant reduction of memory requirements
  • Fixed multiple resource leaks
  • Fixed ~10% performance regression introduced in version 3.1.0
  • Various other small improvements and bug fixes

ARX 3.1.0

  • Implement microaggregation as a new coding model
  • Add support for Oracle DB
  • Fix bug in statistics when used with subsets
  • Fix some resource leaks
  • Various other small improvements and bug fixes

ARX 3.0.0

  • Implemented methods for analyzing re-identification risks (sample uniqueness, average class size, population uniqueness according to different super-population models)
  • Implemented methods for risk-based anonymization (sample uniqueness, average class size, population uniqueness according to different super-population models)
  • Improved reading and writing of CSV files
  • Implemented auto-detection of data types
  • Implemented methods for data cleansing
  • Improved handling of missing values (“null” values)
  • Implemented new visualizations of the solution space
  • Improved query builder
  • Fixed (non-critical) bug in d-presence model. Solution might not have been optimal.
  • Various small improvements and bug fixes

ARX 2.3.2

  • Bugfix: handle zeros correctly in computing the geometric mean for relative comparisons
  • Change license to Apache License, Version 2.0

ARX 2.3.1

  • Various bugfixes
  • Performance improvements

ARX 2.3

Critical bugfixes

  • Correctly enforce d-min in (d-min, d-max)-presence
  • Remove option “protect sensitive associations” due to faulty implementation

Enhancements (ARX Engine)

  • Implement novel pruning strategy
  • Introduce localization for data types
  • Improved subsystem for measuring data utility
  • Implement new default metric (Loss)
  • Implement support for attribute weights
  • Implement configurable coding model for Loss metric
  • Implement configurable aggregate function for most metrics

Enhancements (GUI)

  • Provide tooltips and clipboard support for all tables
  • Improvided data visualization
  • Remove legacy Swing components
  • Support attribute weights and parametrizable coding model
  • Visualize data and hierarchies with virtual tables
  • Updated NatTable to version 1.1.0
  • Fix fonts in tables on OSX

ARX 2.2

  • Update data visualization components
  • Performance improvement for non-monotonic criteria
  • Bugfixes

ARX 2.1

  • Improved data import
  • Support for continuous variables
  • Functional representations of generalization hierarchies
  • Support creating multiple transformations simultaneously
  • Improved support of data types
  • Improved data visualizations
  • Expose all features via API and GUI
  • Various bugfixes

ARX 2.0

  • Implement d-Presence
  • Support for research subsets
  • Support for multiple sensitive attributes
  • Improved algorithms for non-monotonic privacy criteria
  • More flexible API
  • Improved GUI
  • Various bugfixes

ARX 1.0

  • Implement basic k-anonymity
  • Implement distinct-, entropy- and recursive-(c,l)-diversity
  • Support t-closeness with and without hierarchies
  • Initial version of the GUI