Re: Comparison between DITA and S1000D

[john_hunt@us.ibm.com]

Scott and Erik,

I recently participated in a meeting
with a work group (chaired by Tyde Richards) of the IEEE Learning Technology
Standards Committee to compare/contrast DITA with S1000D as potential XML
content models for learning. I can offer a few observations based on that
discussion, which provide some additional background and elaboration on
Erik's response.

First, by way of background, S1000D
is a spec for technical publications initially developed by a European
aviation industry consortium. It's now gaining wider support as a general
spec for content targeted at electronic performance support for military
maintenance applications. 

DITA and S1000D share a number of key
characteristics:

1) Both chunk information into self-contained
reusable units, called "topics" by DITA and "data modules"
by S1000D. 

2) Both can handle specific types of
markup for specific types of information. S1000D information types include
Description, Procedural, Maintenance schedules, Fault Isolation, Crew/operators,
and IPD; DITA core types include concept, task, and reference. 

3) Both provide extensive support for
specifying additional metadata about topics or data modules and specific
elements within them. S1000D provides built-in and extensive numerations
of metadata relevant to aviation and military equipment; DITA provides
extensible metadata modeled after a Dublin core base.

4) Both provide mechanisms for organizing
topics for output to specific deliverable sets. S1000D uses a "publication
module" (PM) to list data modules for publication building; DITA uses
maps, which can structure content hierarchically or by relationship to
other topics. 

As such, both DITA and S1000D support
a topic-based content reuse model, which identifies content by specific
information types, distinguishes individual units according to specific
metadata, and organizes content for delivery with an external map. 

DITA, however, has several key distinguishing
characteristics, compared with S1000D:

1) Information type specialization.
With S1000D, the information types are all predefined and enable specific
support for military and aviation maintenance publications. With DITA information
type specializations, support for new information types can be added quickly,
as specializations of existing types. DITA could potentially model all
of the S1000D information types as specializations of existing DITA types,
for example.  

2) Domain specialization. S1000D includes
specific vocabularies that get used across different information types
to describe specific content domains, such as loading/off-loading procedures,
munitions descriptions, safety, emergency, and so forth. With DITA domain
specialization, the specific domain vocabularies needed to support S1000D
could be modeled as specialized extensions of existing DITA content domains.


3) Specialization with a fallback. In
the case of both topics and domains, the DITA specialization architecture
supports a generalization mechanism, allowing specialized vocabularies
a "fallback" to a common ancestor as needed for exchange, integration,
and reuse with content from other domains or information types. 

4) Processing reuse. DITA specialization
architecture also enables processing reuse, whereby specialized structures
only require specialized processing where needed, and revert to general
processing rules as the default. This makes it possible to quickly add
specialized support for new information types to existing processing models;
support for external content can be brought as a generic DITA type and
use base processing rules. 

DITA thus has the potential to provide
a unifying XML content model, which can enable sharing and reuse of content
across information domains. For example, with DITA, you could integrate
military equipment instructions or aircraft maintenance procedures with
medical or pharmaceutical guidelines.

S1000D does have several key features
that distinguish it from DITA, including:

1) Specification of a Common Source
DataBase (CSDB) mechanism for the storage and retrieval of source content
from a common repository.

2) A process data module, which can
model a scripted procedural flow, including branching, looping, and filtering,
across a set of related content modules. 

All the above said, S1000D presents
a well-developed and extremely detailed and thorough information model
-- the base S1000D spec comes in at over 2,000 printed pages! Modeling
all of the details of S1000D with DITA topic and domain specializations
would be a large task. Erik Hennum's reply points at some useful starting
points for strategizing about how to apply DITA principles to S1000D content,
perhaps to enable interoperability between these two content models.

I'd welcome further thoughts and discussion.

John
___________________________________
John Hunt
IBM DITA Learning Architect
Lotus User Assistance Architect
IBM Software Group/Lotus Software
1 Rogers Street, Cambridge, MA
john_hunt@us.ibm.com