Create an externally accessible Universal Metadata Database, and integrate it into ArcGIS

Idea created by tshindler on May 4, 2018
    Reviewed
    Score40
    • tomi.ronkko_GTK
    • tshindler
    • blythe_spendlove@redlands.edu_univredlands
    • kkoz490

    Metadata needs a universal, externally accessible database, so that all of our data tools can be employed to build, maintain, and access it. This database needs to be integrated into ArcGIS, but accessible to any relational database management system.

     

    Metadata has continued to be a major headache and stumbling block for GIS data stewards. The "standards" involve creating a document that is cumbersome and isolated. It can only be read by a human (or with custom XML code), it cannot be joined to or queried like a database, and can only have limited representations made from it. Editing is a very manual process. So it doesn't get done, and when it does, it is often not used, because the data is only document based.

     

    I propose a that esri create and support a database structure that would contain all the elements required by the various metadata standards, and be stored as accessible table objects in each level of a geodatabase, including SDE, where it could be exposed to all SQL tools. It would be relational, with three core tables:

    • Master Dataset table (containing a record for each feature class, with all the info applicable to the feature class level)
      • Attribute Field table (with a record for every field in all the attribute tables, linked to the Master record
        • Field Value table (containing defined value/description pairs in every field, linked to the field record.) Table domain content would be a subset of this table.

     

    Those three tables would be enough, but the design could include other utility tables, as necessary to assist in such things as facilitating inheritance of metadata from master to derivative datasets, providing tools for an organization to manage the data included in the database, and for the database internal management.

     

    By containing all the metadata in a relational database, its editing could make use of all the wonderful database tools we have (views, models, scripts, reports...), instead of being stuck in a manual one-off document for each dataset, as we currently are. Report formats could easily represent the data in whatever standard form is desired, and easily modified when the standard changes. Each geodatabase could then include its own copies of these three tables, which could allow export and import of metadata to the enterprise in ways that are far more robust than anything we have now. All the power of a SQL database, and related utilities could be brought to bear on the use and maintenance of this data.

     

    Integration with ArcGIS would include auto populating the fields that can come from the system (as currently done), but also, any domains in a geodatabase would automatically populate records in the field value table. Import export tools would be available, and consequently could be included in models and scripts. Imagine having the data manipulation power that we currently have for maps and tables applied to metadata!