Tamr Launches Spring 2019 Release of Flagship Data Unification System

Tamr has announced the general availability of the Spring 2019 release of the company’s patented data unification system. Purpose-built to leverage machine learning, human knowledge, and—where appropriaterules to solve data integration challenges, Tamr enables organizations to create unified data assets that fuel analytic insights and operational improvements.

Initially conceived by Turing Award winner Michael Stonebraker and his co-inventors who published their research in early 2013 about the Data Tamer System for tackling large-scale data curation challenges, Tamr was founded to commercialize the research with initial backing from NEA and Google Ventures.

“Tamr’s Spring 2019 release extends our advantage over traditional data integration tools,” said Mark Marinelli, Tamr’s head of product management.  “The new capabilities in this release delivers a powerful alternative to MDM and ETL for large organizations seeking a faster, more accurate, and cost-effective way to unite, master, and classify data from their many siloed systems.”

New Features in Unify Spring 2019 include:

  • Persistent IDs, which enable a new suite of data mastering workflow features that provide users with enhanced reporting on how clusters of related records change over time, track lineage, and publish results to their downstream systems. New visualization capabilities enable data preparers review cluster changes before publishing master data downstream to apps like PowerBI and Tableau.
  • Golden Records, a core feature introduced to represent the best data available about an entity and where users will turn when they want to ensure they have the correct and complete version of master data for an entity.
  • Data Transformations for Pipelining enhance the curation process by giving users the ability to develop, test, and execute data transformations at scale through the user interface. Along with time and cost savings, the accuracy and usability of the unified datasets are enhanced with transformations making the data invaluable for all downstream applications.
  • Python Client for Tamr Unify, now public on GitHub and for both data scientists who want to produce robust, trusted analytics and IT personnel who seek to automate workflows using a familiar language.
  • Active Learning for Categorization, which supports the building of accurate machine learning powered classification models with a minimum amount of training labels. The Tamr system identifies the highest impact records for humans to review and label as training data, thus delivering the maximum uplift to the accuracy of the model with the least amount of effort.

To learn more about these new enhancements, go to