On Feb 14, 2014, at 3:00 PM, Kingsley Idehen wrote:
An Ontology is (IMHO) very much a kind of Data Dictionary. I wait to be convinced otherwise.
But as you have repeatedly expressed, you/Ontolog/LinkedData has zero interest in legacy systems. Have I missed something?
I think I'm on pretty safe ground that moving legacy data to OpenData is a non-starter. When one removes data from its native locale, it tends to go wonky (that's a technical term).
A little detail is that the contents of a data dictionary isn't what people typically think of as data.
Data is something like the value "12345" in a field in a database.
The data dictionary would contain the definition(s) for "ZipCode". And perhaps, if we're really lucky the rules for valid ZipCode values, but it would likely not contain "12345" as a value for ZipCode. Much less 2M records with ZipCode values.
The person using a database is looking for say how many customers who've spent more than $10,000 last year in ZipCode 12345.
The person using the data dictionary is asking questions about the systems that contain that ZipCode + sales values. Where ZipCode will very likely be used across dozens/100s/1000s of applications under every imaginable name under the sun.
An appropriate use of data dictionary would be to research the potential size of a project to expand ZipCode from 5 digits to 9, 11, or 13. How many data structures, sorts, programs, & reports will need to be changed?
If it existed in the context(s) of the 1000s of applications & artifacts that potentially touch this ZipCode expansion project, I'd like to think a good Ontology might be useful to finger the various labels that represent ZipCode... but so far I haven't seen a glimmer of interest in such a pedestrian task.