Skip to main content

Digital library for computer scientists welcomes participants

ITHACA, N.Y. -- More than 30 academic departments and research laboratories are now participating in the Networked Computer Science Technical Report Library (NCSTRL). The NCSTRL (pronounced "Ancestral") system compiles the technical reports of leading-edge research from computer scientists around the world and provides an efficient, user-friendly method for researchers, students, engineers, and others to access this research over the Internet.

The NCSTRL project has grown rapidly from the experimental digital library efforts undertaken in the Computer Science Technical Reports (CS-TR) project, sponsored in 1992 by the Advanced Research Projects Agency (ARPA) through the Corporation for National Research Initiatives (CNRI), and in the National Science Foundation (NSF)-sponsored WATERS project. The NCSTRL effort continues to evolve, with the hope that it will indeed become the progenitor of future generations of digital libraries.

Technical reports often are the first published notice of the hottest research findings in computer science, predating print publication in conference proceedings or journals by up to two years. Previous efforts to make these reports available electronically were largely on a per-institution basis, with no systematic facility for uniformly searching for or accessing the reports at the separate institutions.

The still evolving NCSTRL project "makes a very large body of breaking computer science research rapidly available to researchers and programmers throughout the world," said Dean B. Krafft, chair of the NCSTRL Working Group and a senior research associate in the Department of Computer Science at Cornell University. "It also serves as a test bed for research in digital libraries."

Indeed, "NCSTRL is not primarily about technology," according to Barry Leiner, associate director of ARPA's Information Technology Office and responsible for the CS-TR project. "It's about creating a new capability in this country and building the organizational and policy framework to support it." Thus, NCSTRL is an evolving effort by a number of institutions to create a digital library infrastructure for computer science documents, not merely another Internet-based collection.

This is not to say that NCSTRL ignores the user. The system provides a seamless interface to servers at many different institutions working cooperatively to provide access to the library materials. "It's transparent to the user. All the reports are available from a single source in a uniform manner," said Carl Lagoze, project leader for the Digital Library Research Group at Cornell's computer science department.

The project can be characterized in three ways. It is an operational networked library, a vehicle for exploring issues of organization and policy and a framework for incorporating new technology. The NCSTRL collaboration will involve combining the efforts of many partners to develop on-going solutions to the technical, organizational and policy issues raised by this project.

A steering committee, with representatives from a number of computer science related organizations, is currently being formed to provide oversight for NCSTRL. Participation in the networked library is welcomed from all computer science research laboratories and Ph.D.-granting computer science departments.

"NCSTRL is an excellent vehicle for carrying out the efforts begun in the CS-TR project," said Robert E. Kahn, president of CNRI. "The CS-TR project had as one of its goals to enable computer science technical reports to be more easily accessible via the Internet."

The CS-TR project enabled the development of several components that are key to the current NCSTRL system. Among these is Dienst, a library access system conceived of by Jim Davis, a Xerox researcher in the Design Research Institute at Cornell, and developed and refined by researchers at Cornell's Computer Science Department and Xerox Corp.'s Wilson Center for Research and Technology; the handle system, a location-independent naming service from CNRI; and a format for communicating bibliographic records over the Internet, originally developed by researchers at the Information Sciences Institute of the University of Southern California and refined by researchers at Stanford University.

NCSTRL also draws from the collections and concepts of the WATERS project. Leaders in the WATERS project include researchers at Old Dominion University, the State University of New York at Buffalo, the University of Virginia and Virginia Tech.

As participation in NCSTRL continues to expand, "NCSTRL will provide the widest possible exposure for research results, while computer scientists in government and industry labs as well as graduate students and even undergraduates will be able to call up research results not yet available in printed form," explained Alan Selman, professor and chair of computer science at SUNY Buffalo and one of the founders of the WATERS project.

Institutions that wish to contribute reports to the library can install NCSTRL software in one of two versions, "standard" and "lite." The standard version supports viewing documents in multiple formats, with customized user interfaces and features such as thumbnail or single page view. The "lite" version offers centralized indexing and imposes simpler software requirements on the participating institution. Both versions support search by author, title, or abstract keyword, and downloading of the reports.

Software for both versions can be downloaded over the network at no cost to the participating institution. Once the software is installed, an author can contribute a formatted paper in a matter of minutes and have his or her latest results in a highly visible location available anywhere on the Internet.