"Living Documents" BOF

Co-Chairs:  Peter Deutsch (peterd@cc.mcgill.ca)                 March, 1992
            Alan Emtage	  (bajan@cc.mcgill.ca)


The Living Documents BOF met Tuesday, March <blat> 7pm-10pm. The
preliminary agenda called for discussion on a wide range of topics
related to the creation and implementation of Living Documents but in
practice the majority of the discussion revolved around data
representation issues for network-based information discovery and
delivery systems.

Much of the discussion centred upon the characteristics needed to
implement a practical scheme for Universal Document Identifiers,
contrasting these with a proposal for Unique Document Serial Numbers.
UDIs have been proposed to allow multiple information systems to
communicate location and access information. Initial proposals that
had been circulated by Tim Berners-Lee, Brewster Kahle and others were
discussed and these were compared to the information needed and
currently provided by such systems as Prospero, WWW, WAIS and others.
No firm conclusions were reached, but it was agreed that a mailing
list (nir@cc.mcgill.ca) would be created to pursue this issue with a
goal of producing a document standardizing UDIs for Internet use.
Initially, all attendees of this BOF are to be placed on the list, and
existance of the list is to be announced to the Internet community.

Discussion concerning Unique Document Serial Numbers centred around the
perceived need to identify and compare the _contents_ (in contrast to
the location) of documents in an internet environment. Ideally, we would
have a means for:

		a) Identifying the contents of a document
		   and comparing it with other documents without
		   copying and comparing them directly.

		b) Identifying derivative works and ancestral links
		   between documents.

		c) Identifying documents that contain the same
		   information despite representational changes
		   that do not add or delete information contents.

It was generally accepted that the first of these could probably be met
with relatively straightforward signature schemes, but that the last two
would be difficult or impossible using strictly syntactic means. At
least one archive site administrator (Mark Baushe "mdb@nsd.3com.com")
has subsequently implemented such an MD5-based signature scheme at his
site (ftp.3com.com) for testing purposes. Details on accessing these
signatures will be posted to the nir@cc.mcgill.ca list.

The discussion continued across a range of topics, examining the other
issues to be addressed in implementing Living Documents and
network-based information systems. The following list was drawn up
outlining some of the issues to be addressed in subsequent work:

Univeral Document Identifiers:
	- design, documentation and deployment. Issues involved include
	  the need to encode individual access methods and specific
	  location information within a specified access method. An
	  initial proposal for such a scheme had been circulated by Tim
	  Berners-Lee prior to the meeting. A copy is available by
	  anonymous FTP from info.cern.ch in the subdirectory

Unique Document Serial Numbers:
	- design, documentation and deployment. Issues involved include
	  identifying specific documents, version control and derivation

Cataloguing Information:
	- Librarians already make use of far more cataloguing
	  information than any of the experimental systems currently 
	  in use on the Internet. Work with those directly involved in
	  library science working with extending MARC records, ISBN
	  and ISSN numbers is called for.
Discovery mechanisms:
	- There remains a large open problem in rapidly and
	  efficiently discovering the existance and location
	  of information in a large distributed computing environment.
	  The proposed UDIs and UDSNs may enable such systems to be built
	  but additional wrk is still needezVCd. There are problems both
	  in locating individual service providers and specific pieces
	  of information.

Authentication and Access Control:
	- Security issues were not discussed in depth, but it was agreed
	  that such issues would become more important as large-scale
	  systems are developed and deployed.

Editorial Control:
	- Again a topic touched upon only briefly, it was suggested by
	  one participant that true Living Document systems would have
	  to include some method of imposing editorial control.

Mailing List:


Mailing List Archive:
	anonymous FTP to:	archives.cc.mcgill.ca
	subdirectory:		"/pub/mailing-lists/nir-archive"