Daniel E. Atkins
The University of Michigan Digital Library Project
and Dean and Professor, School of Information
University of Michigan
Ann Arbor, Michigan
D-Lib Magazine, July/August 1996
This document is a response to four questions D-Lib proposed to the six institutions funded under the current NSF/ARPA/NASA Digital Library Initiative. It is assumed that the reader will have some familiarity with the projects at least at the level of the articles in the May 1996 issue of IEEE Computer Magazine.
1. What is available in the test beds for experimentation among the DLI members?
The University of Michigan Digital Library (UMDL) project currently has six large publishing partners who have contributed over 2,000 titles of content to the project. We are in active negotiation with several more. Some of this material is unlicensed. Through specially arrangements we may be able to provide licensed material for use by other sites.
Use of the ontology
A central issue in UMDL is that of the description of content and terms and conditions for use of information goods and services. We have formalized description methods for information objects and services. These descriptions are used in a metadata structure we call a conspectus. We will share both with others.
Agentware and Other Software
Functionality of the digital library increases through building agents in three categories: interface agents, collection agents, and a variety of mediator agents. The mediator agents form teams to respond to specific information brokering tasks between requests from the interface agents to the collection agents. The agent software library now includes agents such as those to provide access to a Broad System of (search term) Ordering and thesaurus. Other specific agents include task planners, registry agents, remora agent, indexers, and an early version of an auction server. We also have tools for building agents. Other software modules include an enhanced Z39.50 client, the Inquiry Interface developed for use in teaching middle and high school science, and our FTL search engine.
2. What is available in the test beds for experimentation among other groups?
All of the items in question 1, although perhaps on different terms. We are also looking for partnership in build collection interface agents around commercial collections to federate them with UMDL.
3. What are the plans for the test beds at the end of the project?
Broad constituencies of the University of Michigan consider leadership in establishment of global digital libraries and related collaboration environments to be of great strategic importance to the quality and vitality of the University of the future. The NSF/ARPA/NASA UMDL Project is the primary basic research and development project in this area, but is complemented by a variety of other nearer term and more applied projects to provide digital library services to the UM community immediately. Examples can be found at http://www.lib.umich.edu/libhome/electres.html and at http://www.jstor.org/. We have tried to anticipate success.
Technology from the Digital Library Initiative, collections, and experience are already rolling into the new Digital Library Production Capability. This organizations is jointly sponsored by the University Libraries (http://www.lib.umich.edu/), the UM Information Technology Division (http://www.itd.umich.edu/), and the School of Information (http://www.si.umich.edu/). The testbeds is also supporting the agenda of the new Michigan Media Union (http://www.ummu.umich.edu/) and related academic outreach and virtual university experiments.
4. What can be done now to anticipate the continued operation of the test beds (use & R&D) beyond the current project?
The key is to building a successful inter-institutional production digital library operation. We are focusing primarily on serving academic communities but also have experiments underway on the future of public libraries (see the Internet Public Library [http://www.ipl.org/] and Community Networking System [http://www.si.umich.edu/Community/]).
We are supporting inter-institutional work through the National Digital Library Federation involving several major academic libraries and the Library of Congress. The group is targeting partial ways (both technical and organizational challenges) to federate DL collections. One major issue at present is the considerable variability in capability among institutions for federation (e.g., interoperability, metadata, commerce, etc.), and the considerable administrative distance on many campus between the library and computing sectors.
Most of the agreements that we and others have with content providers are for a bounded term with a bounded user community. The long-term success of digital libraries will depend heavily on solving the type of issues we are addressing about licensing and payment mechanisms (pricing models) in the UMDL architecture. Digital library frameworks and environments must provide incentive for continued commercial participation. We are now optimistic about long-term use within the University of Michigan community, but most certainly will not be able to provide our complete content “free to the net.” We are also optimistic that inquiry-based approaches to science education being support by UMDL in the Ann Arbor Schools is being written into the formal school curriculum. The take-up is starting.
Long term viability of a digital library federation requires an incentive for large number of parties to participate and the entry cost for participation to be reasonable.
We are just now approaching a system which is robust enough to explore some of our conjectures, for example scaling (won't know until we try more), incentives for commercial participation, and more advances interfaces. We recognize the challenges of building in context of rapidly changing standards for convergence/coherence. When will there be another Web like explosion? How do you design projects that accommodate massive paradigm shifts?
As exciting as current developments are, we believe that in the DLI we are now building “horseless carriages.” There is a future we can’t yet see or describe well. There are much more fundamental opportunities for new approaches to information and knowledge work environments. What are the new authoring formats, new media, new genres? What will access mean? What if we had agentware as the beginning -- now what can you build?
July 9, 1996
Copyright © 1996 Daniel E. Atkins