1.  User gets a document.  A user somehow has previously obtained a
permanent document id (a list of ways in which that might have
happened is below).  The user gives the permanent document id to the
user client with the intention of examining the document.

The user client sends the permanent document id to a Navigation
service, which returns a URL of an Archive service for this document.
The client then sends the permanent document id to the Archive
service, which returns the property list of this document.  The client
and the user review the property list together, deciding whether or
not the cost and Conditions of use are acceptable, and a suitable form
(representation) is available.  Assuming that both conditions are met,
the client requests part or all of the document from the Archive
service.  The client provides a browser suitable for the document
representation.  The user browses the document.

Scenario 1 is the basis for all scenarios in which a user somehow
obtains a permanent document id and wishes to see the document itself.
Some of these scenarios are:

    - receiving the ID in an advertisement directly from a Publisher
    - receiving the ID in an announcement from an Awareness service
    - finding the ID on a bulletin board
    - receiving the ID in an e-mailed message
    - querying a Discovery service which returns the ID
    - ID learned by following a citation found in some other document
         in its reference list or bibliography

2.  User makes a query.  A user, with the aid of the user client,
formulates a query.  The client sends the query to a list of Discovery
services for which this user has previously indicated a preference.
The Discovery services independently process the query and each
returns a list of permanent document id's.  The client deduplicates
the document id's (perhaps by invoking a third-party deduplication
service not specified here), retrieves bibliographic information about
each remaining document from the Discovery service, and presents this
information to the user for further selection.

The user selects a document from the list and proceeds as in scenario

3.  Citation following (soft).  The user is browsing a document and
notices a citation in the text for another document of interest.  The
user points out the citation in the image to the user client, which
identifies the corresponding text and uses heuristics to formulate a
query that is likely to find the corresponding permanent document ID.
This scenario then proceeds as in scenario two.

4.  Citation following (hard).  Some documents contain permanent
document ID's as part of their citations.  The user points out the
citation in the image to the user client, which picks out the
corresponding permanent document ID from this document.  This scenario
then proceeds as in scenario one.

5.  Adding one document to a collection.  A Librarian
receives an advertisement for a document and, following scenario one
decides that the document is a suitable addition to the collection.
The Librarian's client asks the Archive service for the text
corresponding to the document.  The Librarian's client uses
the result as the basis for indexing this document in this collection.

6.  Tracking an Archive.  A Librarian may decide that the
collection should contain most things stored by a particular Archive
service.  Periodically, the Librarian's client sends a
"What's New?" query to the Archive service, providing as an argument
the transaction number received from this Archive service on the
previous "What's New?" query.  The Archive service responds with a
list of permanent document ID's and the current transaction number.
The Librarian then proceeds as in scenario 5.

Note that a Librarian manager has two ways of discovering things to
add to its collection:  passively awaiting arrival of advertisements
from Publishers and Awareness services, or actively asking "What's
New?" of Archive services.

7.  Publisher creates a document.  A Publisher assembles a document in
the desired form or forms.  The Publisher's client then invokes the
addition method of an Archive service.  The Archive service accepts
the document, creates a permanent, unique document ID for this
document, and returns that ID to the publisher.

The Publisher's client registers the permanent document ID and the
location of the Archive service with a Navigation service.  The
Publisher's client may also optionally register the document with any
number of other identification services (for example, the copyright
office, the Library of Congress, and the issuer of ISBN's), each of
which assigns a name from its name space and associates that name with
the unique document ID in its Navigation service.

Finally, the Publisher's client advertises the existence of the
document by sending messages containing a description and the
permanent document ID (and any other registered identifications)
directly to users and to Awareness services.


Loose ends:

1.  Lists of representations are patched in.

2.  Immutability of Archive didn't seem to matter.

3.  Contracts between Publishers and multiple Archive services.

4.  The responsibility of registering an archive's unique ID follows
a different trajectory from the responsibility of registering other
identities for a document.

5.  The loose end list is not complete.

Change log:

10/14  Add list of kinds of clients
       Relabel Repository service as Archive service.
       Expand description of discovery service to recognize that it
          also can provide bibliographic information.
       Relabel Location service as Navigation service, add method for
          getting data into it.

10/28  Expand publisher's scenario on creating a document to include
          registering the document with any number of identification

Last updated 10/28/94. Converted to HTML 11/2/97.
Return to Library 2000 home page.