SyntheticGenerators

This is not an existing tool, but something I was discussing with Bryan Heidorn and which I think could very easily be developed to assist us with testing the scalability of the SDD schema and the capabilities of any SDD software we develop.

One of the challenges in the SDD schema is the number of Key/KeyRef pairs which need to be handled, along with all of the associated constraint processing. We need to get a better feel for the performance implications of these characteristics. Fortunately it seems likely that these factors are largely independent of the actual meaning of any particular Terminology, Entities or Descriptions elements. We could therefore create a program to generate SDD test documents varying a small number of parameters. The parameters could be something like: NumberOfCharacters, NumberOfStatesPerCharacter, NumberOfClassDescriptions, NumberOfObjectDescriptions, etc. We could vary the parameters to alter the depth of the class hierarchy and the number of specimens per taxon and the complexity of the terminology (along with whether or not it is imported/included). The actual elements could be little more than "State 1006", "Description 214", etc., although we could vary the balance of NaturalLanguageDescriptions and CodedDescriptions. I think this could be a very simple program but could give us a wide variety of different documents against which to test SDD processors.

-- DonaldHobern - 21 May 2004