Classification of Business Documents

1
Classification of Business Documents

• DITA BusDocs Subcommittee Meeting
• 21 January 2008
• Presentation with Notes from the Focus Group
  Meeting of 14 Jan 2008

2
Meeting Summary

• Classification focus group members include Howard
  Schwartz, Eric Severson, Amber Swope, and Michael
  Boses. Howard was not able to attend the meeting
  due to travel
• Michael presented the enclosed PowerPoint as a
  starting point for the discussion
• Discussion was captured and incorporated into the
  PowerPoint under the heading, Notes
• Next steps
• Eric will work on a preliminary mapping of a
  limited number of document types that illustrate
  the mapping
• The focus group will present a summary of what we
  have discussed to the full subcommittee during
  the January 21 meeting

3
Introduction - 1

• The need for a classification system for business
  documents arises from
• The desire to indentify the specific document set
  that is being addressed by the subcommittee, as
  well as the rationale behind that selection
• The ability to further analyze the document set
  using a refinement of the same characteristics
  used to classify them

4
Introduction - 2

• What type of characteristics are important?
• Documents can be classified in many ways. The
  most common way used is a semantic classification
  based upon the textual content of the document
• The subcommittee approach is different since we
  want to classify documents based upon their
  structural characteristics since it is the
  structure of business document that will need to
  be harmonized with DITA

5
Potential Structural Characteristicto Consider
when Classifying

• Is it a narrative?
• Narrative complexity
• Document length
• Tree depth
• Tree balance
• Table frequency
• Table complexity
• Graphic frequency
• XML vocabularies
• Transclusions
• Notes Eric feels that repetitive structures will
  be an important characteristic
• Amber suggests that whether a document references
  external system data might be important as well
• Howard Understanding the business purpose might
  be important as a characteristic.
• Eric could be interesting but maybe not the
  driver
• We will capture the information as part of the
  analysis
• Ann Its possibly a different level of
  classification
• Josef translation should not in itself change
  the structure, but perhaps what we want to look
  at is documents with variants in them.
• Howard--Business documents will have different
  challenges than technical publications

6

• Higher level model
• Structures that are not linked to semantics that
  can then be correlated to documents for different
  usage
• The end-game is to say where does DITA fit in?
• semantic neutral way of classifying
• Apply the general to specific usages later
• Eric concept, task, and reference were
  specializations to begin withare they even
  meaningful for business documents?
• Howard-- Informational, vs. persuasive? Intent or
  purposesdoes it correlate to structuredoes it
  dictate structure, does it matter for reuse?

7
First-level Classification

• Notes while the concept is good, none of us is
  happy with the terminology. In particular, we
  need to come up with an alternative for Forms.
• The purpose of this slide is to say that there
  are business documents that are out-of-scope.
  This is our first level?

8
Form-Narrative Scale
Subject Document

• Metric
• Ratio of total elements to total words
• Notes Eric What is a form? How do we keep from
  excluding documents with structures that we need
  to address, because we called a form? Something
  to describe form that isnt based upon its
  implementation. XML blurs the distinction
  between documents and data
• A Elements are structural in nature. We need
  to define what type of elements we will use to
  arrive at the ratio

9
Most Significant Characteristic?

• Once we have established that it is a narrative
  document, what is the next most significant
  characteristic to examine?
• Notes, general agreement with the presentation,
  that it would be the tree depth of the document

10

• Eric- DITA is trying to apply best practices to
  writing is this a fundamental thing about
  writing or is it just tech pubs?
• Should there be a more generic task that could be
  specialized into a tech pubs task?
• Ann- what we have now is a specialization for
  tech docs and so it fitsit is possible to start
  at higher at a more generalized level
• Interesting that paragraphs have topic
  sentence. The topic sentence may be an important
  bridge that allows us to introduce the concept of
  topic based authoring to the business community
• Business documents are maturingare tech docs
  more mature? Tech docs are most often not read
  for pleasure and are random access information
• Writing for reuse has a significant impact on how
  content is writtendoes it invalidate some of our
  common business document structures?

11

• Types of reuse
• The ability to flow one persons content into
  another persons content and have it hold up
  contextually
• The ability to have content presented as a result
  of a query or aggregation and have it hold its
  integrity as a single unit of information
• Will the message change depending upon how
  someone arrived at iteither in the original
  context or by itself?
• All this ties back to the maturity model that
  will help organizations move to a best practice
  approach to authoring. This will give us
  something valuable for business and acceptable to
  the DITA community.
• Now our classification can also correlate to this
  issue.

12
The Need to Quantify Hierarchy

• The author of the highly nested document is using
  structure to communicate semantics.
• Hierarchical Scale
• Ratio of total transitions in hierarchy to total
  elements
• Notes General agreement. No specific comments

13
Qualifying Narrative Density

• Narrative Density Scale
• Average paragraph length for paragraphs gt 100
  characters
• Notes no specific comments

14
Recap of Characteristic Importance

• Is it a Narrative?
• Narrative complexity
• Document length
• Tree depth
• Tree balance
• Table frequency
• Table complexity
• Graphic frequency
• XML vocabularies
• Transclusions
• Notes Eric- we need to address repetitive
  structures (i.e., topics) and constrained
  structures. What do repetitive structures and
  constrained structures mean to DITA?
• Michael the number of paragraphs per section
  seems importantbut what is a section?

15
Notes Additional Discussion

• Discussion of an SOP as it relates to repeating
  structures
• One approach to an SOP is for it to be very
  verbose, with only 4-5 structures
• Another approach is for it to be very terse, with
  20 structures that add semantics to the content.
• The goal of XML in general when applied to
  narrative documents, is to imply more and more of
  the semantics through the document structure
• Document linearity with repeating structures as
  a structural characteristic provides random
  access to the information in the document.
• Repetitive structures appear to be as important a
  characteristic as the tree depth, if not more.
  Repetitive structures to a degree indicate
  whether the document is a reference or something
  intended to be read end-to-end?
• Repetitive structures cause a document to
  actually be a collection of mini-documents, each
  that could stand alone