Notes on Controlled Vocabularies

2014-10-23 Joachim Wackerow (meeting  with Knut Wenzig, Justin Lynch, Sanda Ionescu)

Current Approach of DDI Controlled Vocabularies

• Value of the Code
• Descriptive Term of the Code
• Definition of the Code

Hierarchy is expressed by a separator (“.”) in the code. Leftmost string is the highest level.

Both, term and definition can be multi-lingual.

The American English expression is the canonical version.

Current CVs in DDI 3

• External controlled vocabularies
• CodeList/CategoryScheme with nested codes for expressing hierarchy
• Enumerated lists in XML Schema

Multiple Purposes

• Code lists
• Classification
• Controlled vocabularies

Requirements

• General
  • Ideally only one type of structure for multiple purposes
  • Usage of existing structures preferred if possible (like SKOS in the Semantic Web)
  • Same structure for internal representation in the model/specification and external representation. Reasoning: easy processing, same software solution, only different reference.
  • Validation of structure, keys, and possibly values
    • What should be validated: keys, values, relationship, dependency, requirement, …
    • XML Schema: what should be validated by XML parser, what in a secondary-level validation
• Simple approach
  • Hierarchy
  • Multi-lingual text for term and definition
  • As simple as possible, easy to process
• Complex
  • Requirement of items
  • Relationship of items
    • use case thesaurus
    • CV for keys, CV for values of key/value pair

Conclusion

Current sense: Two different approaches required: a simple CV structure similar to the approach for the current DDI Controlled Vocabularies and a more complex approach for additional requirements like validation and defining requirements.

Ideally the complex approach could make use of the simple approach.