How Five Industries Will Benefit From the Grove Paradigm

1
How Five Industries Will Benefit From the Grove
Paradigm

• Steven R. Newcomb
• srn_at_techno.com
• TechnoTeacher, Inc.
• HyTime Track, XML Europe 1999, Granada, Spain,
  April 29

2
How this talk is not the same as the paper...

• The paper in the Proceedings is organized by
  industries.
• This talk is organized around the features of the
  grove paradigm that are useful to those
  industries.
• Much has happened since the paper was written.
  There are more than five industries involved in
  grove-based projects.

3
First, whats a grove? Some definitions

• A set of interconnected nodes conforming to
  ISO/IEC 107441997 Annex A.4.
• Ready-to-run objects. The application-internal
  information that is being created, edited, or
  used by an application.
• The form of information that is the result of
  parsing an interchange document (written in XML,
  for example).
• Information that is present in an interchange
  document, but which, for application programming,
  is structured differently in the grove. The
  abstract API to the information set.

4
But thats the DOM, right?

• No. The DOM is an API. A grove is a set of
  nodes -- no methods.
• But its true that an API to a grove of an XML
  document could serve the same purposes as the
  DOM.
• And the DOM is an example of an interface to an
  XML grove.
• Groves are more powerful, general and
  sophisticated.

5
Use by reference

• Groves allow any component(s) of any resource(s)
  expressed in any notation(s) (XML or non-XML) to
  be transcluded into any components of any
  resources expressed in any notations (XML or
  non-XML),
• ...without changing any of the resources
  involved.
• Groves allow there to be a single authoritative
  copy of each datum, regardless of its notation.
• Groves allow each datum to be expressed in the
  most cost-effective notation.

6
Linking

• Groves allow any component(s) of any resource(s)
  expressed in any notation(s) (XML or non-XML) to
  be hyperlinked to any components of any resources
  expressed in any notations (XML or non-XML),
• ...without changing any of the resources
  involved.

7
Links are valuable assets

• Groves allow the links to be exported as
  standards-conforming, non-proprietary XML
  documents.
• Specifically, groves allow addresses to be
  expressed in conformance with standards.
  Standards-conforming software from any vendor can
  resolve the addresses.

8
Management of links and uses-by-reference

• Groves provide knowledge of all anchors of all
  links, and of the links of all anchors,
  regardless of the notations of the anchors.
• Groves provide knowledge of all the sources of
  the transclusions in each document, and of all
  view documents into which any given transcluded
  material is transcluded, regardless of the
  notations of such sources.

9
Management of links and uses-by-reference

• Groves allow links and uses-by-reference to be
  classified via inheritable common architectures,
  as well as by the application.
• Groves greatly facilitate applications that
  provide traversal and transclusion services.
• Groves allow the side effects of changing an
  information asset to be predicted.

10
Technology independence

• Groves can support software that creates and
  maintains links, bookmarks, annotations, auditing
  records, etc. that cannot be lost merely because
  of subsequent changes in computer technology, or
  changes in system vendors.
• All such referencing information can be stored in
  freestanding XML documents that contain addresses
  expressed in a way that is correctly resolvable
  by any grove-aware application.

11
Bookmarks

• Groves allow the notes of caseworker
  professionals, expressed in XML, about particular
  casefiles can incorporate bookmarks to components
  of materials in any notation, without changing
  the materials.
• Such notes can be used to provide continuity of
  corporate behavior toward (and understandings
  with) customers, without changing the casefiles
  themselves.

12
Shop notes

• Groves allow shop notes to be maintained by shop
  floor technicians in documents that are separate
  from the manuals.
• The notes can be made to appear as if they were
  actually part of the documents (or not).
• Subsequent versions of the manuals can have the
  same shop notes, without having to update all of
  them.

13
Scalability

• Lightweight in-memory grove-based applications
  are practical in PDAs.
• Lightweight ISAMs can support PC-oriented grove
  applications.
• Heavyweight object databases can support large
  repositories of information resources.
• Mammoth grove-based systems are supportable with
  multiple databases. The links expressed by
  resources within them can address anything in any
  database, thus tying everything together into
  what is, in effect, a single gigantic database.

14
Topic maps

• "The GPS of the Web" -- Charles F. Goldfarb.
• Topic maps are supportable without groves, but
  not in their full generality and not at their
  full power.
• Topic maps are insupportable without independent
  linking. (It's very hard to support independent
  linking without groves, and very easy with
  groves.)

15
Workflow

• Groves allow everything about workflow,
  regardless of the notations of the produced work,
  to be driven and controlled by XML documents
  containing annotating hyperlinks.
• The components of information resources that need
  review, editing, rework, approval, or
  re-publication, or for which proposed changes
  exist, can be appropriately annotated, without
  changing them,
• regardless of the notations in which the
  resources are expressed, or the manner in which
  they are stored.

16
Programmer productivity

• The grove paradigm elegantly fits all
  applications involving the addressing of
  arbitrary information components.
• Every node is as addressable as every other node.
• A node can be anything
• from a whole document to a single character,
• from a movie to a pixel,
• from a complete engineering drawing to a single
  polygon, line, or point.

17
Programmer productivity

• Even incoming ephemeral information, such as the
  information arriving from the diagnostic
  connection harness of a piece of equipment that
  is being repaired, can be a grove that is as
  useful as any other grove.
• Databases can be made to appear to be groves.
• Metadata can be grove-ified. (Example WebDAV.)

18
Notation processors

• A Notation Processor can be the application's
  mature and highly capable partner for chores
  common to applications that use and/or manipulate
  data that conforms to
• a notation,
• a class of information objects, or
• a database schema.

19
Auditability of transactions

• Groves allow transactions involving
• creating, changing, and deleting content
• creating, changing, and deleting links to content
• re-using (and ceasing to re-use) content
• ...to be as auditable as desired.

20
Locally controlled DTDs

• Groves allow applications to support inheritance
  of information architectures (meta-DTDs).
• Meta-DTD-specific property sets mean re-usable
  software modules, so
• Groves reduce the cost of software
• Groves increase the uniformity of interpretation
  of data by applications.
• This is vital in e-commerce involving technical
  product descriptions.

21

• A demonstration of a grove management technology.

22
DOM vs. Grove

• The main difference is that the DOM is not
  schema-driven. It has a single (implicit and
  as-yet-undocumented) schema -- a schema for the
  XML notation.
• The DOM can only be used for documents expressed
  in XML, whereas any information, in any notation,
  can be represented as a grove.
• Even in XML-land, the DOM does not provide access
  to the information sets conveyed by specific XML
  document types, whereas a grove can be a set of
  nodes that reflect the information set, and that
  may bear little resemblance to the interchange
  form of the same information.

23
What about the XML Information Set?

• The W3C's XML Information Set project is a great
  thing, and it is urgently needed in order to
  allow the addresses of XML document components to
  be interchangeable.
• If the XML Information Set were expressed as a
  Property Set, then it would be a description of
  a grove -- an XML grove. In any case, it will be
  very similar to a property set for XML.
• The XML Information Set is expected to be a
  description of the features of XML, but not
  described in a fashion that could be used for
  other notations.

24
The Mainstream is More than XML

• There will always be plenty of notations other
  than XML.
• Their components must be addressable so that they
  can participate in the mainstream of information
  processing.
• Groves allow that to happen.

25
XML DOM solutions are for XML

• XML DOM solutions dont allow information to be
  expressed in any form but XML. All information
  has to be converted into XML.
• Conversion is expensive. If information changes,
  it's a recurring expense.
• Conversion takes time, so the XML version is not
  always up to date.
• What if you have created links and other
  references to the XML form, and the original
  changes?
• What if you have changed the XML form, and need
  the changes to appear in the original form?
  Reconversion may be even more expensive!

26
Groves No conversions necessary.

• Grove-based solutions allow the information to
  stay in its original form, without requiring it
  to be converted into XML.
• This is essential, for example, where products
  are being produced, delivered, maintained,
  supported, redesigned, redocumented, and
  delivered simultaneously.
• All five industries, and many others, share this
  requirement.

27
How does the Grove Paradigm work? (1)

• For a given notation, the interesting aspects are
  named in a schema called a property set.
• For example, in a property set for photographs,
  we might say, People's noses must be
  addressable. Lets make sure all the noses
  become nodes in the grove that is made from
  photographs. We'll call such nodes 'Nose
  Nodes'.
• So, its like writing a DTD for information that
  isn't in XML you decide whats interesting about
  some set of information, and you give each such
  thing a property name, which is like creating an
  element type name.

28
How does the Grove Paradigm work? (2)

• When you create a grove, everything that was
  regarded as interesting
• exists in the grove (and nothing that wasnt
  interesting is there), and
• in effect, every component says what it is. Its
  like tagging, but without the tags.

29
If I Have XML, the DOM, etc., Why Groves?

• You need to be able to talk about things other
  than XML.
• Talking about things means being able to point
  at them, so you can say what youre talking
  about.
• If you never need to talk about anything that
  isnt already in XML form, THEN YOU DONT NEED
  GROVES!
• The reverse is also true.