BITS: The New Kid on the Block

If we want to tag books in XML – why BITS?

Why not:

• Docbook
• DITA
• TEI

... all of which have been used successfully for books in the past ...?

(We leave aside the central question: what is a “book”?)

Scope of coverage: fit the tag set to the domain

• DITA: hyperlinked, indexed resources in structured prose (technical documentation)
• Docbook: technical monographs (esp. computer software)
• TEI ... ?! (academic scholarship and publishing)
• BITS ... !? (commercial publishing?)

Which tag set should I use?

The best answer may depend on which culture you feel at home in.

But this doesn't always determine.

TEI - Text Encoding Initiative

• Since 1987 (SGML-based, then XML)
• Intended to deal with issues of data longevity and portability for scholarship in the (academic) humanities

• Many TEI resources and projects have been developed

  • Scholarly critical editions
  • Anthologies of poetry and prose
  • Dictionaries
  • Manuscript and epigraphic transcription
  • Linguistic corpora
  • Etc. etc.

Difficulties with TEI

(For organizations building publishing systems)

... both too much, and too little, at the same time?

• Over 550 element types (and counting)
• Many very specialized
• Metadata is underspecified for many purposes
• Element semantics can be too loose or too tight

• Schemas must be customized to be useful

  (So my TEI is not the same as yours)

• Architecture is complex

  — So a good customization is not trivial

• Same thing goes for tools as for the schema

On the other hand...

The TEI community has one thing the JATS community doesn't have much of (yet) ...

Experience with heterogeneity

• Users come from many different directions
• With different problems and goals
• TEI offers a unified framework to accommodate them
• Moreover, public TEI profiles or customizations serve particular needs within the larger community
• Some TEI subset will align with BITS

Enter “Best Practices for TEI in Libraries”

See Best Practices for TEI in Libraries: A guide for mass digitization, automated workflows, and promotion of interoperability with XML using the TEI

(Version 3.0, October 2011. See http://www.tei-c.org/SIG/Libraries/teiinlibraries/)

• Addresses needs for organizations such as libraries

• Proposes five levels of encoding from “lightweight” on up:

  1. Fully automated conversion

     (Lights-out conversion from plain text)

  2. Minimal encoding

     Very lightweight encoding of text, may link to page images

  3. Simple analysis

     Paragraphs, tables, lists, graphics

  4. Basic content analysis

     Names, numbers, dates, abbreviations, more complex structures (e.g. drama)

  5. Full analytic scholarly encoding

     (In effect, anything TEI allows)

The best part: these levels are codified

• A schema is proposed for each of them

  (Except level 5, effectively full TEI)

• This means that element and attribute types can be effectively mapped

• And (generally) inferences can be made respecting tagging semantics and usage

Offering stable points of reference for aligning with BITS

Proposal: formally map BITS to TEI and back again

• BITS will correspond to something between levels 3 and 4 of “TEI Best Practices for Libraries”

• A systematic and comprehensive two-way semantic mapping:

  • Will expose commonalities and congruences

  • Will also expose stress points where things are different

    (in available semantics, granularity)

    • Clarifies choices and tradeoffs
    • Suggests where adjustments and improvements are possible
  • Provides the basis for (semi-) automated crosswalks (transformations)

Benefits of this exercise

• Clarify where BITS vs TEI is (not) an either/or
• Feed back into tag set design/maintenance
• (Each community gains from observing the perspective of the other)
• Capitalize on investments already made
• Forestall future duplication(s) of this effort
• Ease development and data interchange going forward

Feedback wanted

• Is this as obvious as it sounds?

  (Any possible reasons not to do this?)

• How should this work be organized and published?

• How can it be motivated and funded?

Code: When do I use BITS? When do I use TEI, Docbook or DITA?

Yes, all of these can potentially do anything that any of the others can do.

And has its particular strengths.

Choosing your tag set: Technical considerations

• Content modeling

  • Docbook comes with elements like KeyCode, GUIButton, FuncDef
  • TEI comes with elements for verse, drama, letters, MS transcription, oral history ...
  • DITA comes with an infrastructure (“maps”) for relating resources and generating hypertext reflecting this organization

• Look especially at

  • Metadata
  • References and citations

• Appropriate level of complexity

  • Be concerned about noise (stuff you won't use)
• Available tools and toolkits

Choosing your tag set: Non-technical considerations

• What do your staff, vendors, partners already know?

  (Already knowing JATS gives BITS a big plus)

• Where can you find support and expertise?