From:       Jerry Saltzer and Greg Anderson, MIT

To:         Bob Kahn, CNRI

Subject:    Continuation proposal


1.  Introduction

The contract between MIT and CNRI is scheduled to end on June 30,
1995.  As we near the end of the originally-contemplated contract
period, we find that we have accomplished many, but not all, of the
original objectives, and we have spent much, but not all, of the
authorized funds.  We believe that with some additional time, and
within the original budget, it will be possible to make substantial
additional progress.  We therefore propose that the contract be
extended, at no extra cost, through December 31, 1995.

There are two reasons why the project has not moved as rapidly as we
had hoped.  First, the original proposal was for a three-year
investigation starting in July 1, 1992, but negotiations on contract
details delayed its formal start until mid-November, 1992.  Although
CNRI did provide a modest advance to allow work to get started before
then, MIT policies prevented us from undertaking the commitments
required for full-scale activities until the contract was actually
signed.  Second, a number of ramp-up tasks proved to be more difficult
than we envisioned.  This combination leaves us with the opportunity
to now continue making progress without exceeding the original budget.


2.  Work completed to date.

In overview, we have developed a scanning station and a scanning
workflow, and have steadily increased the volume of work that that
workflow can handle.  We have scanned about 17,000 pages of Technical
Reports of the two M.I.T. computer science laboratories, and are in
the process of placing those scanned images online.  We have created a
testbed digital library system with a search engine and several user
interfaces, we have populated it with the records of the LCS/AI
reading room and links to the scanned images of the participants of
the CS-TR project, and we have placed it in public use at M. I. T.  We
have completed several academic theses, published some papers, and
participated in architectural discussions of the CS-TR project.  More
detail on the work completed to date will be found in the annual
reports and the quarterly progress reports that have been filed
regularly since the contract was signed.


3.  Work to be done in the continuation.

a.  Increasing the size of the scanned-image TR library.  As indicated
in our latest quarterly progress report, we have been increasing the
rate at which we are scanning technical reports, and as that rate
increases we are developing strategies for overcoming bottlenecks in
handling the large quantities of data involved.  This work will not
only lead to a larger corpus of material on-line, it will also
contribute to our understanding of the workflow issues involved.  In
addition, we will continue to work out ways to automate the
acquisition of PostScript sources of newly-released technical reports
to produce on-line images.

b.  On-line replication system.  We have developed most of the ideas
for, and part of the implementation of, a novel archival replication
system for digital libraries that we believe has considerable merit.
We propose to finish the implementation and place the replication
system into service to gain some experience with it.

c.  Transfer of service responsibility and technology.  One of the
original goals of working jointly with the M. I. T. Library System was
that both the ideas developed during the project and the library
materials collected should have a transfer path into long-term
production systems.  We have already transferred our scanning workflow
expertise into a production environment that we expect to continue to
operate indefinitely, in the Document Services branch of the
M. I. T. Library System.  We propose to work on transferring to the
Library System the responsibility for the continued archiving and
public availability of the collection of scanned images.  We also
intend to continue our active program of transferring technology and
expertise to the M. I. T. Library groups responsible for systems
design.  Since the Library's system design group has embarked on a
joint project with GEAC to develop an advanced, client/server library
system, we believe that this is a unique opportunity to influence the
direction of at least one widely-used product.

d.  Writing things up.  We have been focusing mostly on design,
implementation, and production, so by now quite a number of papers and
technical reports providing implementation details and insights gained
from operation can be written.  We propose that this writing be a
major activity of the continuation period.

e.  Continued joint participation.  To the extent that CNRI finds it
useful and our schedules permit, we will continue to participate in
both on-line discussions and meetings called by CNRI of the CS-TR
participants and the Distributed Library Initiative participants.

4.  Personnel.

As you know, Jerry Saltzer is retiring from the M. I. T. faculty on
July 1, to a part-time role as Senior Lecturer.  This role allows him
to continue as a principal investigator, and he will be continuing to
work with Greg Anderson to guide the project.  All other staff
continue to be available and currently intend to continue to work on
the project.  In addition, we anticipate that some of the students who
have been working with the project during the last year will continue
to do so in the fall.