Warning: Creating default object from empty value in /home/johnnz/public_html/wp-content/themes/simplicity/functions/admin-hooks.php on line 160

The Role of Metadata in Data Quality

3 Responses to “The Role of Metadata in Data Quality”

  1. Richard Ordowich September 11, 2013 2:25 pm #

    Metadata is not data about data

    We see this description for metadata and it is not really helpful. Metadata is the attempt to objectively describe data. Since data is subjective to begin with, metadata is needed to describe the data in a way that (almost) everyone agrees with.

    We being with the technical metadata like structure and the value domain and then we get into the challenging metadata like name and definition. Names can be resolved using naming conventions and a controlled vocabulary. Definitions tend to be more difficult since they are people’s understanding, interpretation or perception of the data.

    We then require semantic metadata, but once again it is possible to define the semantics using objective semantic criteria such as time, place etc. Finally the context of use metadata. Where and how the data used in the organization helps bring out the subjectivity surrounding the data.

    Once you have the technical metadata along with the name, definition, semantics and contexts of use, you have reduced the subjectivity of the data and hopefully developed a consensus of understanding of the data. This is metadata.

    Following these principles for metadata will help improve data quality. Most data quality issues stem from poor; subjective, incomplete and ill defined metadata.

    • John Owens September 12, 2013 12:44 am #

      Hi Richard

      Metadata is and always has been data that describes the format and structure of data.

      In order to refute this definition, you must first provide another to replace it. Nobody has yet done this. Without such a definition it is quite meaningless to then talk about ‘technical metadata’ and ‘semantic metadata’. Adding adjectives to an undefined term does not give it meaning.

      Far from what you assert, this definition for metadata is very helpful. If data analysts and commentators were to acquaint themselves with it and adhere to it, then we would have far fewer people stumbling about in the fog of confusion that has been created by ignorance of its existence.

      So, definition first, arguments against existing definition after that.


    • David Eddy September 17, 2013 12:19 pm #

      Richard -

      > Names can be resolved using naming conventions and a controlled vocabulary.

      Do you find organizations do have controlled vocabularies to support their naming conventions?

      Or do projects just make them up as the projects happen along?

      I—and I’d say this is a BIG if—there is a controlled vocabulary available, how do different projects (or even the same project) resolve the reality of multiple meanings for the same term?

      Dear ol’ CUSTOMER is, of course, the prime example of many meanings.

Leave a Reply to Richard Ordowich