Collaboration, KR, Automatability
Historical
A statement from Tim Berners-Lee, 28 Feb 1995:
There is a sequence of development which may not be very high on
the agenda for commerce on the Internet but may have as important an impact
on the nature of human use of networks. These elements may be tackled
independenty but there is a certain logic to tackling them in order:
- Interactivity: front end update, annotation
- Information structuring
- Knowledge Representation
- Intelligent agents and the transmission of executable code
Introduction
Interactivity
The web was initially designed as an interactive hypertext system
for communication nthrough shared knowledge. However, the lack of
interactive wysiwyg hypertext editors
(apart from the original application) constrained the web to be
a largely "broadcast" medium as fas as its use has gone.
Although the architecture allows for decentralised editing
and publication, the bottlenecks of HTML production tools and
server administrators have made spontaneous linking and
interaction impossible.
The benefits to be gained from raising the functionality of the
web, both clients and servers, to allow interaction will include:
- Allowing groups to use the web at an informal level,
mixing formal and informal hypertext, as a powerful tool for
solving problems and working in teams;
- Allowing authorized people to update and comment on extant information,
reducing the amount of out of date information and hence raising the
overall quality of the web;
- Reducing the administrative overhead of those whose part in the
prduction cycle is largeley mechanical
- Providing though human interaction
a more liveley, stimulating and enriching environment
for all web activities including education, research and commerce.
Information Structuring
There have long been calls for ways of better structuring information
on the web. One part of the solution to this is to allow the small
hypertext nodes of which the web is composed to be linked together
with relationships such as subsections, footnotes, previous versions,
etc.
Once this metainformation is defined, clients will be able
to brovide a richer interface. For example, buttons will be able to appear to
navigate the document structure; the status of documents will be
able to affect their style, color, etc; Graphical maps used for navigation aids
will be able to use different representations to express the structure;
and programs will be able to treat aglomerations of related documents
as groups.
Knowledge representation
As a step toward making a machine-interpretable web of knowledge,
which is an important long-term goal, we must separate statements
which are part of the knowledge base itself from metatinformation about
documents.
Knowledge in the web may be divided ino knowledge which is stored in
the structure of the web, in its typed links, and knowledge which
is stored in nodes written a form of knowledge representation language.
The former allows commonality and interworkingbetween different applications
as it involved standards for relationship semantics. The latter is
more flexible, provides greater functionaloity and less interworking, and
is largely a question of using existing flexibility of the content-type
negotiation system to introduce existing langauges.
In this case some work may be needed to connect the existing langauges into the
web environment (as with
general SGML on the web)
Intelligent Agents
This is an area which is more atthe reserach stage, although
some interesting products may be evaluated.
The areas in which W3C may be involved would typicallybe in defining a common
execution environment for mobile executable objects, and in defining
inter-language communication standards in order for elements in different
langauages to interact seamlessly.
Requirements
Here are some outline strawman points.
For interactivity
- Front-end creation and modification of documents
- Common generic model for user interaction
For Information structuring
- Defined link types for information structuring
- Recommended browser representations of related operations
For Knowledge Representation
- A form for expressing information about the
object described by a document rather than the document itself
- A set of basic link relationship values
- Prototype example web analysis and processing engines
Products
For interactivity
- Reference code library for clients allowing PUT, POST,
LINK and UNLINK (or equivalent)
operations to be easily included
- A sample server allowing authenticated remote update
- A sample client allowing authenticated remote update
For basic information structuring
- A sample client with an interface enhanced by the use
of the basic link types
- A simple tool allowing processing (e.g. printing)
of structured document systems
For basic knowledge representation
- A sample client allowing semantic relationships to be
set up;
- A sample toolkit allowing processing of a semantic web.
Current Situation
The current libwww library has some of the PUT and POST work
underway.
A set of link types has been defined since 1989 but there
has been a confusion between meta
information about information, and information
about other things.
Next Step
A W3C meeting in Sept 11-12
in Boston is to be arranged
to gather input for a refinement of the requirements.