|
Carnegie Mellon University
Report to the Digital Library Federation
October 2004
DLF Newsletters, Volume 5:1 (2004)
A. Collections
Size: 16,098 pages
Scope: A Pittsburgh architecture journal
System: DIVA, WolfPack
Migration: With other DIVA resources
Intellectual property: Permission
Funding creation: Library resources
Sustainability: Library resources
Use: Recently released; 97,161 pages served so far in
2004; active; metadata harvesting and copying into Million Book
Project would increase use
Challenges: Coated paper, graphics, multi-volume
serial
Size: 2,610 pages
Scope: Carnegie Museum of Natural History archival
material
System: DIVA
Migration: With other DIVA resources
Intellectual property: Permission
Funding creation: IMLS funding
Sustainability: Library resources
Use: 6,589 pages served since 2001
Challenges: Sustaining collaboration with another
institution; need critical mass of materials to make collection
more useful
Size: 1,114,105 archival pages and growing
Scope: Papers of distinguished Carnegie Mellon
affiliates
System: Original HELIOS (Heinz-funded software that
became commercial) replaced with DIVA (commercial software
configured locally); some functionalities lacking in new
software
Migration: Already moved once
Intellectual property: Permission
Funding creation: Heinz endowments, family gifts, IMLS
grant
Sustainability: Two family endowments
Use: 90,735 pages served since 2001; metadata harvesting
would increase use
Challenges: Beginning project, different materials
Size: 124,000 volumes (summer 2004)
Scope: Aall public domain materials; India, China, and
other major international partners
System: Mini Universal Library (locally-customized
commercial software)
Migration: Mirrored worldwide and migrated using
post-doctorate fellow Dr. John Ockerbloom's typed object model
techniques for translating data into new formats
Intellectual property: Public domain and permission
Funding creation: NSF, India and China governments
Sustainability: Negotiating with OCLC
Use: Active
Challenges: Feeding international scanning centers with
content, copyright clearance, and others.
Size: 313,413 color book pages and
archival correspondence files
Scope: Books and documents
collected by Henry Posner Sr.
System: DIVA, WolfPack (locally-customized commercial
software to perform data conversion tasks and post-processing
quality control)
Migration: With other DIVA resources
Intellectual property: Public domain and permission
Funding creation: Posner family gift
Sustainability: Posner Fine Arts Foundation if
required
Use: 59,827 pages viewed and 10,380 covers viewed
Jan-June 2004; metadata harvesting and copying into Million Book
Project would increase use
Challenges: Rare, fragile color materials
Size: 584 images to date
Scope: Rare book room collection
System: FileMaker database access via the web
Migration: With other FileMaker databases
Intellectual property: Fair use low image dpi
Funding creation: Donor gift
Sustainability: Library resources
Use: 63,666 images viewed since 2002
Challenges: Library absorbs future costs to photograph
collection additions each year (20-30 posters); posters are large
and difficult to photograph; cataloging is challenging and
time-consuming.
Technical Reports: Public access
through library catalog, http://cameo.library.cmu.edu
Size: 82,103 pages
Scope: Technical reports produced at Carnegie
Mellon
System: WolfPack; accessed through Cameo (library
catalog)
Migration: With other digital collections
Intellectual property: Public domain and permission
Funding creation: Library resources
Sustainability: Library resources
Use: Recently released
Challenges: Identifying materials in public domain; no
OCR
B. Services
Automated Resource Finder, a locally
developed tool, assists the user in finding relevant materials in
a variety of formats and at a variety of levels beginning with a
basic level. ARF is customized so that for specific resources
such as the library catalog (Cameo) or the Britannica Online,
searches are already pasted into the resource search window. This
tool is a starting point for a user. Links to “Ask a
Librarian” are part of the implementation. Consultants,
using data from user testing, recently revised the design of the
ARF interface. Carnegie Mellon usage statistics indicate that the
majority of online catalog and database use occurs by remote
access. Without librarians to guide them, remote users often use
inappropriate materials (indexed by popular search engines) or
become confused and overwhelmed by the many e-resources provided
by the library. ARF helps address these problems. ARF pages were
hit almost 3000 times per month between January and September
2004.
Electronic Reserves: Restricted
access (current students/faculty) in library catalog, http://cameo.library.cmu.edu
Size: 89,281 pages
Scope: Changing set of materials for use in current
courses
System: Locally-developed programs; accessed through Cameo
(library catalog)
Migration: Created for one-semester use only
Intellectual property: Fair use and permission
Funding creation: Library resources
Sustainability: Library resources
Use: 151,506 pages served in 2003 (compared to 6,857
traditional reserves uses)
Challenges: New emphasis on multimedia
Onsite Authentication
Until recently library workstations were
available to walk-in users with no authentication required.
Beginning in spring 2005, all users of public library machines
will be required to authenticate. A 24-hour ID and password will
be available at the circulation desk. For specific research
needs, a one week pass will be available. [Details for onsite
authentication are not yet finalized.]
Remote Access
The library, along with the university, has
grappled with the problem of remote user access. The library
provided access to remote users using a proxy server. The use of
this technology to access library resources was satisfactory part
of the time but left some users dissatisfied and unable to access
needed resources. We now use the IP extension service, also known
as Virtual Private Network (VPN), to connect remote users to
library resources. The library also participated in beta testing
of the Shibboleth software, specifically with JSTOR and
FirstSearch.
C. Systems
Digital Information Versatile Archive (DIVA)
DIVA provides web access to many of Carnegie
Mellon’s digital collections. It uses commercial database
software and open standards to provide a common framework for
digital projects. For example, DIVA’s full-text searching
is provided by Oracle, interface customization is done using Java
Server Pages, and metadata harvesting is provided through an OAI
interface. DIVA allows students and researchers to search,
browse, view and print digital images of books, journals,
technical reports, and archival documents. With specifications
developed by Carnegie Mellon librarians and archivists, DIVA
provides conventional access to library and archival materials,
and adds new functions for searching and retrieving documents,
supporting multimedia, and customizing the structure and
presentation of collections.
Publishers Database
This locally-created interface supports
Copyright Permission Research. See Projects update below.
DIVA Scan
This locally-created interface enables
scanning operators to digitize materials quickly into TIFF
format, create or add metadata, and make initial decisions about
the need for professional processing by archivists.
WolfPack Technology
This suite of locally-customized commercial
software performs large data conversion tasks in a distributed
manner. The WolfPack framework allows the best off-the-shelf
commercial conversion programs to be used in an automated system.
For scaleability, it runs the conversions in parallel on a large
number of machines. WolfPack performs image cropping, de-skewing,
de-speckling and OCR, and creates JPEG and Acrobat files from
scanned images. Built-in quality control mechanisms ensure the
integrity of the digital collection.
DOI Server
This locally-built system supports metadata
harvesting. We use the Digital Object Identifier (DOI) standard
to create persistent URLs for objects in our digital library. A
server was assigned and specialized software was written to
maintain a database of DOIs for our digital projects. This system
allows the physical location of an item to vary throughout its
lifetime while its identifier remains unchanged.
OAI Layer
To make our digital library useful to the
widest audience, we support standards that allow remote tools to
harvest metadata about our collections. Using Open Archive
Initiative (OAI) metadata harvesting protocol in our digital
library software allows content to be indexed remotely. OAI is
scheduled for fall 2004 implementation.
A. Projects
New Project Announcements
Jewish Serials Project
This project will digitize and make available three serials
(periodicals) emanating from the Jewish community in Pittsburgh,
Pennsylvania, 1895 to present. This digital collection will be
valuable to historians, genealogists and religious scholars.
Shull Papers
Early in 2003, the University Archives received the papers
of Clifford G. Shull, who shared the Nobel Prize for Physics in
1994. The American Institute of Physics has provided an $8,000
grant to process the papers and create a finding aid. The Shull
family is providing additional funding to digitize and make
available the collection (41 linear feet of materials
representing Shull's work as a student, researcher and faculty
member at MIT). Shull did his undergraduate work at Carnegie
Tech. This collection will be an important resource for
physicists and those doing research in the history of
science.
Strategic Partnership with FAO for the Million Book
Project
On October 14-16, 2004, Dr. Anton Mangstl, Director of
Library and Documentation Systems Division of the United Nation'
s Food and Agriculture Organization (FAO), and John Reid, Chief
of Technical Services, met with the Million Book Project to
discuss a strategic partnership. On October 16, Dr. Jacques
Drouf, Director General of FAO, and Dr. Charles H.
Riemenschneider, Director of FAO' s North American Liaison
Office, joined the ongoing meetings. FAO produces documents in
five languages-English, French, Spanish, Arabic, and Chinese. FAO
and Carnegie Mellon want to develop a strategic partnership
around several issues-content for the Million Book Project,
enhanced provision of online question answering services, and
placement of PCtvts in three different countries. Action items
from these meetings include:
- FAO will begin to ship duplicate copies of
about 20,000 pages of content for scanning in India. After that
is completed, materials that must be returned will be shipped for
scanning.
- Servers will be shipped to Rome for the
scanning of materials too rare to ship. Eric Burns and Ed Walters
will stop in Rome on their way to India to configure servers and
to consult with FAO technical staff.
- A work plan for the next three years will
be developed with the assumption that FAO and Carnegie Mellon
will cover their own costs in achieving its goals.
- A workshop will be convened in Rome on
February 9-12, 2005, to pull together individuals who can discuss
how to bring FAO information into a service analogous to Google
Answers or Ask Jeeves. A group of experts in different kinds of
online reference services and in the technology that supports
them will be convened.
- The Director General will provide a list of
selected countries and villages for an experiment with
PCtvt.
Synthesis by Morgan and Claypool
Carnegie Mellon will participate in
Morgan & Claypool’s Developmental Partner Program (DPP)
which is designed to provide feedback from a selected group of 30
libraries worldwide on the next generation of digital products.
Synthesis is an innovative product comprised of
“Lectures” which are short documents (75-100 pages)
born digital, that provide a current summary of a research topic
in electrical engineering and computer science written by leading
researchers in their field. Founded by Michael Morgan (who
previously founded Morgan Kaufman) and Joel Claypool (who created
the CRC Press Engineering Handbook Series), the company is not
burdened with legacy systems and is free to develop new user
oriented products and business models.
Update on Existing Projects
Copyright Permission Research
The Libraries began conducting research on acquiring
non-exclusive copyright permission to digitize and provide open
access to books in 1999. An initial study was conducted to
determine the likelihood of publishers granting permission to
digitize and provide free-to-read web access to their copyrighted
books. A second study sought permission to digitize and provide
open access to the copyrighted books and archival materials in
the Posner Memorial Collection. A third study seeks permission to
digitize and include out-of-print, in-copyright books in the
Million Book Collection. Lessons learned from each study are
applied in the next study in an effort to decrease the
transaction cost and increase the success rate. For example,
copyright permission efforts in the Million Book Project are
focusing on books published by scholarly associations and
university presses because the feasibility study and the Posner
project revealed that they are more likely to respond and grant
permission than commercial publishers.
University Libraries’ research has begun to establish best
practices for copyright permissions:
-
Determining whether a book published
1923-1963 is still in copyright. Using a re-keyed version of the
copyright renewal records prepared by Distributed Proofreaders,
Million Book Project partner Michael Lesk has developed a way to
submit queries from a batch of library catalog records and
determine which books are in still in copyright.
-
Identifying and locating copyright holders.
Publishers often merge or go out of business or copyright reverts
to the author or the author’s estate. The U.S. Office of
Copyright retains no records of copyright ownership. Often
multiple letters are sent to different addresses seeking
permission for the same title. Sometimes the copyright owner
cannot be located. Carnegie Mellon legal counsel advised us to
assume “permission denied” if the copyright holder
cannot be found.
- Reducing the cost of acquiring copyright
permission. Copyright permission work in the Posner study, not
counting administrator time or legal consultation fees, cost
approximately $78 per volume for permission granted. Consequently
an entirely different approach is being used in the Million Book
Project (MBP). Rather than requesting permission for designated
titles, the MBP is requesting permission to digitize and provide
open access to all of the publisher’s out-of-print,
in-copyright books. Using this approach, we have received
permission to include roughly 47,000 copyrighted books in the
Million Book Project at an estimated cost of $0.42 per
book.
- Funding the copyright permission work. We
want to include 500,000 copyrighted titles in the Million Book
Project. A grant proposal that would have provided financial
support failed.
- Increasing the response and success rates.
Our research indicates that engaging publishers in a dialog and
offering to give them copies of the digitized books increase the
response and success rates. Seeking permission for older,
out-of-print materials and providing a web site where publishers
can see the quality of the scanned books are also helpful
strategies.
Challenges: Partnering with others
to conduct further research that would ultimately reduce the cost
and increase the success of acquiring copyright permission to
digitize and provide open access to books.
Digital Image Database
At the beginning of the fall 2004 semester, the University
Libraries introduced to the campus community its Digital Image
Database. The database consists of images requested by Carnegie
Mellon University faculty and students as well as images
purchased through commercial vendors. Images are made accessible
via the web for study and use in Carnegie Mellon University
classroom teaching and presentation. With the introduction of the
Digital Image Database, we hope to improve authorized access to
our image collection as well as eliminate contention typically
associated with individual 35 mm slides.
Each image is scanned as the highest possible resolution TIFF
from which various sized JPGs are derived (constraining tool will
automatically adjust height and width). The sizes available are
Thumbnail (128 x 99 dpi), Working (640 x 480 dpi), and
Presentation (1024 x 768 dpi). Once the JPGs are generated the
archival TIFF is moved to magnetic storage. Image resolutions
were based on recommendations/standards from the Library of
Congress (American Memory Collection) website, ARLIS conference
presentations, various listservs and professional
communiqués.
The intellectual content provided to catalog the images will be
generated by library staff or, in the case of purchased
collections, by the vendor. The Getty Research Institute’s
Vocabulary Databases, made available via the web to support
limited research and cataloging efforts, may be consulted to
standardize creator names, confirm dates, etc. The Image Database
will be available to Carnegie Mellon users only through Internet
Protocol authentication. Copyright and/or source information will
be provided within the image's record.
Million Book Project
The project’s vision is to bring one million books free
to read to the internet. NSF has provided $3.6 million in funding
for equipment and travel with China providing $4.8 million for
research and scanning and India $25 million for a broad range of
language initiatives including scanning.. The collection will be
a collection of collections of out of copyright, public domain,
and permission granted materials. Several university presses and
scholarly associations have given permission to have their lists
scanned, or are negotiating their permissions. Indigenous
materials in India and China are being scanned and U.S. books are
being shipped to and from international scanning centers.
Research objectives for the project include security, copyright,
digital rights management, optical character recognition
accuracy, OCR of non-Roman languages and scripts, automatic
metadata creation, summarization, intelligent indexing, machine
translation, storage formats, and search engines. Library
challenges: Identifying collections to be sent to
international scanning centers, ensuring metadata standards,
locating books with permissions given, and finding a
sustainer.
B. Programs
None at this time.
In 2004, Carnegie Mellon University
Libraries completed the work outlined in our Digital Library
Plan, 2001-2007 (available below). Our current challenge is
to create a new Digital Library Plan for the university.
return to top >>
|
Copyright © 2004 Digital Library Federation. All rights reserved.
Last updated:
|