M87systems Corporation Logo

M87 Galaxy

M87 Galaxy

Metadata Don'ts

Many corporations start a Technical Metadata initiative as an after thought to a major data warehouse/business intelligence implementation and fall prey to the common Metadata Don'ts.

Don't build it yourself

A strategic metadata initiative is far too complex to build it yourself. Many first time initiatives treat metadata like a standard cataloging application and don't include the flexibility required to handle the multitude of scenarios a metadata store can require.

Don't treat metadata like a Gadget, It's a Solution

A strategic metadata solution should be designed to address real corporate issues. The metadata implementation should always stay focused on the problem to be solved.

Don't rely on People to Populate it

Metadata store maintenance should be a fully automated process whenever possible. Depending on people to keep it green will only lead to bad, or out-of-date metadata, and bad decisions.

If you do need to manually populate and maintain metadata, make sure you have a well defined process to keep it up-to-date.

Don't fall into a Metadata Black Hole

Metadata black holes are metadata content that:

  • Are never used
  • Cannot be used
  • Was easy to capture, so we did
  • Does not provide the intended benefit
  • Was populated but almost impossible to keep green

When deciding what metadata to capture, populate and maintain, the following points should always be considered:

  • Who is the intended audience
  • What to capture
  • How to capture it
  • Where to capture it from
  • How to recognize change
  • Who is responsible
  • Why do we want it
  • When and how will it be updated
  • How will it be accessed (navigation)

Don't automate Descriptive Content

Many organizations use CASE tools to generate their data warehouse loads. The CASE tool is used by developers to encode the business rules to populate the data store. Often these transformation, and cleansing rules are extracted from the tool to populate the metadata store. For the most part, these transformation rules are totally unreadable by your business users, and serve only to confuse and drive them away from your solution.

Don't ignore the standards

There are many existing metadata standards specifically designed to address certain kinds of metadata problems:

  • Dublin Core for describing electronic documents
  • CWM, (Common Warehouse Model), for data warehouse information exchange
  • IEEE-LOM, (Learning Object Metadata), for technology supported learning
  • CSDGM, (Content Standard for Digital Geospatial Metadata), for digital geospatial metadata

Use the standards for maximum interoperability, and standardization.