W3C Push Workshop, 8-9 September 1997

[Other Papers] [Briefing Package]

Title: Common Datacast Architecture (CDA) and the prototype "Channel Builder"

Speaker:  Michel J. Hebert, MITRE Corporation


THE SCENARIO: Its a typical staff meeting.  Toward the end of the meeting,
John, my boss, asks me what I know about "extranets."  At first I think its
just another buzz word for intranets, then I remember reading an article
about a technology that is being developed to allow organizations to share
information, without compromising their security.  John explains that a goal
of his research program is to find effective and efficient ways to share
information with our sponsors, while working with existing firewall technology.
Further, he thinks that this type of technology could possibly help us solve
some sticky problems we are having with telecomuting.  John has instructed me
to find information on these subjects and keep it coming for the next 3 to 4

YESTERDAY'S SOLUTION:  I indicate that I'll get on the web and surf around to
see what I can collect on the subject by this afternoon.  Also, I'll contact
the library and set up a continuous literature search, where we can receive
excerpts from the various trade journals on a weekly basis.  I  provide the
library with a list of web sites, related technologies, and other key words 
that will help them track down the required information.

TODAY'S SOLUTION: I indicate that we should contact the MITRE library and
have them put together a channel based on extranets for broadcast throughout the
corporation.  I  provide the library with a list of web sites, related
technologies, and other key words that will help them track down the required
information.  The MITRE library uses the Channel Builder software provided by
my team to create and name the channel, connect the channel to various
hunter/gatherer agents, and connect the channel to various webcasters (e.g.,
PointCast, Backweb) and an email listserv for delivery of the information. 

TOMORROW'S SOLUTION:  Since John has been running the Adaptive Personal
Channel (APC) software on his workstation, he doesn't even come to me for the
information.  The APC software has been monitoring John's email, bookmarks,
web surfing expeditions, and files saved on his work station to determine what he
is interested in.  The APC realizes that John has been interested in extranets
and asks John if he would like to add this to his personal channel. 
On John's suggestion, his profile is changed and the information is sent to 
his local Channel Builder, the requests are dispatched to the appropriate agents and
the information is broadcast to John via PointCast.  Because John recognizes that
extranet information would be useful to the rest of the corporation and our
sponsors, he instructs the APC to establish a general extranet channel,
selecting the channel for specific release to our sponsors.

There are two primary ideas explored in the proposed presentation:

1) In the wake of a very young market with little standardization, and
information providers required to generate content in several proprietary
formats, another layer in the "push architecture" which collects, aggregates,
and publishes information to the various push products would be useful.  We
discuss a generalized architecture called the Common Datacast Architecture
(CDA) and the prototype "Channel Builder" work going on at MITRE.

2) Looking forward to the time when the number of channels rivals the number
of web sites, the ultimate problem of finding information that is of use to the
typical internet user still remains.  One solution lies in the direction of
agents that watch what you do (the email you save, the web sites you
bookmark, the documents you store,...) and then adapts you personal information 
profile and therefore, your personal channel, to the information you need now. 
We discuss our research, architecture, and prototype work for an Adaptive
Personal Channel (APC).