WInte.r version 1.4

Added new similarity measures:

• TF-IDF cosine similarity of strings
• Similarity of geo coordinates

Added a new matching rule with missing value penalty:

• A linear combination matching rule which applies an explicit penalty for missing attribute values

Extended result logging features:

• Examine data fusion results on record level
• Improved user notifications for easing error analysis

Further changes:

• Report blocking time during identity resolution
• Publish winter on maven central
• Illustrated the usage of the new features in the tutorial

WInte.r version 1.3.2

Changed logging framework to slf4j and minor bugfixes.

WInte.r version 1.3.1

Fixed:

• Fixed an error in the logging system for matching that occurred in combination with parallel processing

WInte.r version 1.3

Added new logging and debugging features for:

• Blocking
• Matching
• Data Fusion

Added new facade classes for data normalisation:

• The ValueNormaliser can be used to transform string values into various data types and interpret units of measurement
• The DataSetNormaliser can detect data types and transform complete datasets into a normalised format

WekaMatchingRule:

• Added support for external models (e.g. learned with RapidMiner) using the PMML format
• Added access to the model's description and performance after training
• Improved export of generated features for external model learning

Extended Web Tables data model:

• Added join
• Improved projection
• Improved support for functional dependencies and candidate keys

Further changes:

• Improved parallel processing in ParallelProcessable
• Added framework extension: Metanome Integration
• Added a detailed tutorial on how to use WInte.r for data integration

WInte.r version 1.2

Performance improvements, a new blocker and several bug-fixes.

WInte.r version 1.1

Replaced Yoda Time with Java Time. Not compatible with version 1.0!

WInte.r version 1.0

WInte.r has come! First release of the WInte.r Framework.