Does DITA need tags?

CIO Enterprise Content Center of Excellence | ©Copyright IBM Corp. 2014, 2015 1
Does DITA Need Tags?
Michael Priestley, Enterprise Content Technology Strategist, IBM CIO
Lu Ai, Program Manager, Content Standards and Structure, IBM Marketing
Carlos Evia, Director of Technical and Professional Writing, Virginia Tech
Content owners: Michael Priestley, Lu Ai, Carlos Evia

• Why lightweight DITA?
• Authoring DITA in markdown
• Publishing DITA with JSON
• Next steps
Does DITA need tags?

Lightweight DITA

Collaboration and integration depend on standards
4

Barriers to collaboration
Process
Lack of governance
Lack of feedback
Culture
Lack of incentives for reuse
Incentives for reinvention
Content
Inconsistent content types
Inconsistent classification
Technology
Lifecycle silos tie authoring to delivery
Formats tie content to authoring system

Can you get there with DITA?
• Yes.
• But…
• Challenges of complexity and format-dependency
• Opportunity:
• Lightweight DITA as a new OASIS standard (in development)

When enough is too much
Full DITA
Lightweight
DITA
“Here’s
something
simple!”
“We need
more
features!”
“It’s too
complex!”

Lightweight DITA vs Full DITA
Full DITA Lightweight DITA
Topics ~100 elements ~30 elements
Maps 10 elements
(+30 shared with topic)
3 elements
(+3 shared with topic)
<p>What elements are
allowed in a paragraph?</p>
dl parml fig syntaxdiagram imagemap image
lines lq note hazardstatement object ol pre
codeblock msgblock screen simpletable sl table
ul boolean cite keyword apiname option
parmname cmdname msgnum varname wintitle ph b
i sup sub tt u codeph synph filepath msgph
systemoutput userinput menucascade uicontrol q
term abbreviated-form tm xref state data data-
about foreign unknown draft-comment fn
indextermref indexterm required-cleanup
image
ph (phrase)
b (bold)
i (italic)
u (underline)
sup (superscript)
sub (subscript)
xref (link)
data

Lightweight DITA at OASIS
Current focus on:
• Industry scenarios
oMedical
oSoftware development
oEducation
oMachine industries
oMarketing/ecommerce
• Format mappings
oXML
oHTML5
oMarkdown
oJSON

Collaboration and integration depend on standards
1

Authoring DITA in Markdown

Mapping DITA to HTML5 (HDITA)

Mapping DITA to Markdown (MarkDITA)

• The “beautiful structure” of a step
• When the DTD is away…
o… the “steps” will play
1. Am I a step?
<step>
<cmd>Or am I a step?</cmd>
</step>
Overprotected child vs. the free spirit
<!ENTITY % step.content
"((%note;)*,
%cmd;,
(%choices; |
%choicetable; |
%info; |
%itemgroup; |
%stepxmp; |
%substeps; |
%tutorialinfo;)*,
(%stepresult;)? )"
>

The Markdown wars
• There is no Markdown standard
• Well, there’s CommonMark
• And there’s Gruber’s original
• And many other flavors

• Should we go the Fountain way?
• Or should we embrace the simplest
scenario?
# I am a topic
This is my abstract or shortdesc
- A point
- Another point
MarkDITA is not a new flavor

• {#identifier .class}
• The best/only way to talk about headings
• A solution to the <section> problem
# I am a topic
With a shortdesc paragraph
## And I am a section {.section}
Add a scoop of Pandoc header attributes

# The point of it all
I can sum it up here
I can say some more stuff
## Stuff {.section}
And so on
- This
- Is
- A List
From < > to ##

# How to do something {.task}
Introduction to something
## Prerequisites {.prereq}
Find some time to do it
## Context {.context}
Be prepared to do it
1. Plan it
2. Do it
Even better

{.section}
{.example}
{.prereq}
{.context}
{.result}
{.postreq}
@outputclass equivalents

• [key] works for keyref declared in a map
• conref… that’s another story
• Unless we go the GitHub way:
{{ site.data.conrefs.phrases.reusable }}
Content reuse

# Topic collection {.map}
- [First concept](c-first.md)
- [First task](t-first.md)
- [Second task](t-second.md)
Maps
Or
<map>
<title>Topic collection</title>
<topicref href="c-first.md"
format="markdown" />
<topicref href="t-second.dita" />
</map>

My relationship with Jarno Elovirta

• But we already have MkDocs and Jekyll
• And Pandoc transforms to EPUB and
Why bother with DITA?
Next slide: DITA benefits and advantages

DITA benefits and advantages

Publishing DITA with JSON

JavaScript Object Notation (http://json.org)
What is JSON
Data type
• Number
• String
• Boolean
• Object
• Array
• Null
Data structure
• name/value pair (NVP)
"name": "Lu Ai"
• object
{"firstname": "Lu","lastname": "Ai"}
• array
“presenters": [
{
"firstname": "Lu",
"lastname": "Ai"
},
{
"firstname": "Michael",
"lastname": “Priestley"
}
]

Why JSON
• Lightweight, easy to parse format
• popular alternative to XML as a data interchange format
Use case at IBM
• Dynamic content delivery via API
CMS1
CMS2
CMS3
Enterprise
Content
Catalog
JSON
Converters
json
xml
csv
DITA
user2
user
1
json

IBM Case Study Structured Content Pilot
• Component based
content management
• Open standard
• Reusable
• Measurable
• PDF only asset
management
• Proprietary data
model
• Low reusability
• Difficult to track and
measure
Current State Future State
DITA

IBM Case Study DITA Specialization: a non-semantic approach
topic
title: macys’.com: …
section: {overview}
dlentry: {the need}
dlentry: {the solution}
dlentry: {the result}
…
section: {client background}
section: {need}
section: {solution}
section: {solution components}
section: {benefit}
….
This less constrained
specialization approach
provides flexibility, but does
not offer much semantics.

IBM Case Study DITA: a mechanical translation to JSON
DITA JSON
• Property name has minimal semantics
• Unnatural/complicated structure in JSON
• Difficult to query

IBM Case Study DITA Specialization: a semantic approach
This tightly constrained
specialization approach
provides rich semantics, but
is less flexible.
casestudy
title: macy’s.com
csprolog
metadata
clientbackground
csneed
title
shortdescription
needbody
cssolution
title
shortdescription
solutionbody
solutioncomponents
transformation
csbenefit
title
shortdescription
benefitbody
longquote
shortquote
csinsidestory
…

IBM Case Study DITA: a semantic approach
DITA Design
• Adopt a semantic approach in
developing Case Study DITA
specialization
• Allow minimal formatting tags in
Case Study DITA
• Analyze specific content needs
and use cases for case studies to
best determine when to use
semantic and formatting tags

IBM Case Study DITA: semantic translation to JSON
JSON Design
• Flat structure in JSON whenever
possible
• Structured hierarchy is only
used if it makes semantic sense
• Formatting tags removed or
retained in JSON “value”
• Program friendly: simpler
structure, easier queries,
cleaner output

Query Comparison: compile an “overview” based on two different JSON data models
Non-semantic Semantic
Complex query
Section[@.title=“Overview”].dl.dlentry[?(@.dt=“The need”)].dd
+
Section[@.title=“Overview”].dl.dlentry[?(@dt=“The solution”)].dd
+
section{@.title=“Overview”].dl.dlentry[?(@dt=“The result”)].dd.q
+
section{@.title=“Overview”].dl.dlentry [?(@dt=“The result”)].ph
Simpler query
csneed.shortdesc+cssolution.shortdesc+csbenefit.shortdesc

IBM Case Study JSON Data Model
• industry, country, language, category, product,
collateraltype, productgroup, creator, sme, owner,
reviewdate, expirationdate, contact, businesspartner
metadata
• title, subtitle, clientinfo, clientcontact, need, solution,
benefit, clientexperience, quotes, solutioncomponents,
formoreinfo, copyrightstatement, disclaimers
content
• casestudy_pdf, clientlog, industryimage, ibmcolorblock,
videoassociatedassets

Next Steps
• Finalize Case Study DITA specialization and JSON data model
• Case Study JSON Schema
• DITA2JSON converter

Next steps/resources

• Continuing design work at the Lightweight DITA subcommittee:
ohttps://www.oasis-open.org/committees/tc_home.php?wg_abbrev=dita-
lightweight-dita
• Public discussion at the LinkedIn group:
ohttps://www.linkedin.com/groups/Lightweight-DITA-4943862/about
• Test out early prototype mappings:
ohttps://github.com/jelovirt/dita-ot-markdown
ohttp://any2dita.com/
What’s happening

Does DITA need tags?

More Related Content

What's hot (19)

Similar to Does DITA need tags? (20)

More from Michael Priestley (8)

Recently uploaded (20)

Does DITA need tags?