Re: [dita] Comparison between DITA and UDEF

[Don Day <dond@us.ibm.com>]

Erik Hennum has tried to post this response via the Comment Form and had no luck, so I am posting his researched reply here on his behalf:

As I read the initiative, UDEF (Universal Data Element Framework) defines global categories for data. These categories can be associated with elements by means of a standard attribute. By equating elements in distinct vocabularies, UDEF has a primary goal of enabling data interchange (especially EDI).

In essence, the attribute asserts

objectType.propertyType is the category of this content

The categories themselves are organized in a kind of taxonomy. That is, UDEF defines a restricted set of object and property "classes" that constitute the top-level categories for the taxonomy, which can be extended by "qualifiers" registered with a central service. For instance, the "Customer Person Last Name" type would have an object type of the Customer qualifier on the Person class and a property type of the Last Name qualifier on the Name class. UDEF assigns a number to each class and, on the shared repository, an alphanumeric identifier to each qualifier, so the attribute can express the category quite tersely as the "as.5_5.10" GUID.

In making semantic assertions about machine-processable data, UDEF is in the same problem space as TopicMaps or RDF. These general-purpose alternatives support a much greater range of assertions (not just categorizing content but qualifying relationships between resources) as well as supporting decentralized definitions. That's not to disparage UDEF but merely to acknowledge its deliberate simplification to encourage faster implementation and adoption. The fact that the UDEF definitions and assertions could easily be represented in RDF, for instance, suggests that an organization could adopt UDEF now and later transition to RDF later.

DITA has a different problem space: modelling human-readable discourse as granular, extensible content objects. The DITA concerns are with the structure of discourse, with specialized structures for special-purpose discourse, and with managing relationships between content objects. In short, DITA is not about identifying equivalent, isolated values in arbitrary locations within different content but, instead, about structuring discourse with semantic precision for maximal readability, navigability, processability, and extensibility.

As with TopicsMaps and RDF, UDEF could complement DITA by enabling additional semantic processing of DITA content. In particular, DITA elements with the same data as other vocabularies could store a GUID value in the otherprops attribute. For instance, the otherprops attribute of the author element in the topic prolog might store a GUID value. One alternative would be for a transform to match the element name and add a GUID attribute while generating XML for processing with UDEF tools. Another alternative would be for the shell DTD or Schema to add a fixed GUID attribute to selected elements. This attribute would break interoperability with other DITA adopters who have not added the GUID attribute, but such a trivial, isolated deviation from the standard could be managed by a transform that strips out the GUID attribute when exchanging content with other DITA adopters.

For Erik Hennum,

Regards,
--
Don Day <dond@us.ibm.com>
Chair, OASIS DITA Technical Committee
IBM Lead DITA Architect
11501 Burnet Rd., MS 9037D018, Austin TX 78758
Ph. 512-838-8550   (T/L 678-8550)

"Where is the wisdom we have lost in knowledge?
Where is the knowledge we have lost in information?"
        --T.S. Eliot