Machine Learning Data Catalogs

Makes sense, also connecting the data to its actual meaning,  semantic ontologies,  is also good to do in the same place.      It should be a broader aspect of governance.   It is also an important fundamental aspect of interpretability to know where the data is coming from, what its stability and credibility are.

How the Machine Learning Catalogs Stack Up  
Alex Woodie in Datanami

You can’t do anything with data – let alone use it for machine learning – if you don’t know where it is. In the age of big data, this is not a trivial matter. It is also the main driver that’s propelling the rise of machine learning data catalogs, which the analysts at Forrester recently ranked and sorted. Just a word of warning: the name at the top of the list might surprise you.

According to Michelle Goetz’s June 21 Forrester Wave report, the percentage of analytic decision makers managing more than 1 petabyte of data (either structured, semi-structured, or unstructured) has essentially tripled from 2016 to 2017. That rapid growth has exposed all manner of problems in company’s existing data management and analytic endeavors.

Two of the biggest challenges that companies face today, Goetz writes, are gathering and managing data in a governed manner on the one hand, and managing the business processes that surround the data analytics activities on the other.

“For EA [enterprise analytics] professionals, relying on people and manual processes to provision, manage, and govern data simply does not scale,” the Forrester analyst writes. “Enterprises are waking up to this fact and turning to data catalogs to democratize access to data, enable tribal data knowledge to curate information, apply data policies, and activate all data for business value quickly.”  .... "