Internet Research: Electronic Network Applications and Policy, vol. 5, no. 2, 1995, pp. 25-36.The NASA Technical Report ServerMichael L. Nelson, Gretchen L. GottlichNASA Langley Research Center, Hampton, VirginiaDavid J. Bianco, Sharon S. PaulsonComputer Sciences Corporation/NASA Langley Research Center, Hampton, VirginiaRobert L. Binkley, Yvonne D. KelloggNASA Dryden Flight Research Center, Edwards, CaliforniaChris J. BeaumontComputer Sciences Corporation/NASA Ames Research Center, Moffett Field, CaliforniaRobert B. SchmunkNASA Goddard Institute of Space Studies/National Research Council, New York, New YorkMichael J. Kurtz, Alberto AccomazziSmithsonian Astrophysical Observatory, Cambridge, MassachusettsOmar SyedNASA Lewis Research Center, Cleveland, OhioAbstractThe National Aeronautics and Space Act of 1958 established NASA and charged it to "providefor the widest practicable and appropriate dissemination of information concerning [.] itsactivities and the results thereof." The search for innovative methods to distribute NASA'sinformation lead a grass-roots team to create the NASA Technical Report Server (NTRS), whichuses the World Wide Web and other popular Internet-based information systems as searchengines. The NTRS is an inter-center effort which provides uniform access to various distributedpublication servers residing on the Internet. Users have immediate desktop access to technicalpublications from NASA centers and institutes. The NTRS is comprised of several units, someconstructed especially for inclusion in NTRS, and others that are existing NASA publicationservices that NTRS reuses. This paper presents the NTRS architecture, usage metrics, and thelessons learned while implementing and maintaining the service. The NTRS is largely constructedwith freely available software running on existing hardware. NTRS builds upon existing hardwareand software, and the resulting additional exposure for the body of literature contained ensuresthat NASA's institutional knowledge base will continue to receive the widest practicable andappropriate dissemination.

IntroductionThe National Aeronautics and Space Act of 1958 established NASA and charged it to "providefor the widest practicable and appropriate dissemination of information concerning [.] itsactivities and the results thereof." To meet this goal, researchers at various NASA installationshave developed several new methods of distributing information to the nation's research andindustrial sectors. One key method is the NASA Technical Report Server (NTRS). The NTRS isan inter-center effort to provide uniform access to various distributed publication servers residingon the Internet. It currently provides access to documents from ten different NASA organizationsspanning the United States: Langley Research Center (LaRC), Dryden Flight Research Center(DFRC), Numerical Aerodynamic Simulation Division (NAS) of NASA Ames Research Center(ARC), Goddard Institute for Space Studies (GISS), Institute for Computer Applications inScience and Engineering (ICASE) at LaRC, the Selected Current Aerospace Notices (SCAN) andRECON databases maintained by the NASA Scientific and Technical Information (STI) Program,the Study of Electronic Literature in Astronomical Research (STELAR) Project from GoddardSpace Flight Center (GSFC), and the Astrophysics Data System (ADS) Abstract Service at theSmithsonian Astrophysical Observatory. All other NASA centers have committed to NTRS andwill be joining shortly.The NTRS is accessible via the World Wide Web (WWW) (Berners-Lee, 1992), a multi-protocolInternet information system, using software such as the freely available and highly popular NCSAMosaic (Andreessen, 1994), developed at the National Center for Supercomputing Applications(NCSA) at the University of Illinois Urbana-Champaign. There are many different WWW"browsers" available, most of them free of charge for all popular platforms (UNIX, Mac, PC,VMS). Figure 1 shows the NTRS page as seen through a WWW client. Implementing NTRS withthe WWW reduced the development time necessary. WWW is used for many services, thus NTRSis built on the tools and lessons learned from many other WWW projects. From the users'perspective, NTRS is one of many available WWW services, allowing access with a consistentand well-known interface. NTRS is available at the following uniform resource locator SFigure 1: NTRS as seen through an X Window System WWW browserProject GoalThe goal of NTRS is to provide "one-stop-shopping" for NASA technical publications. Theintended audiences are researchers and scientists, not necessarily information specialists orlibrarians. NTRS is not the final word in searching and indexing; its intent is to provide maximumconnectivity and exposure to the already existing body of NASA electronic technical literature.

NTRS is an electronic document reuse effort. Many researchers prepare their conference papers,technical reports, and journal articles using sophisticated word processing and desktop publishingtools. High quality document preparation systems for personal computers and workstations haveenabled researchers to produce fully (or mostly) electronic publications. After the paper is printedand sent to the publisher, it is then assembled and preserved in hardcopy format. While there areadvantages to preserving the document in this format, most researchers maintain an electroniccopy of the publication. NTRS attempts to gather the diffuse collection of electronic publications,index them, and offer them to the scientific community.NTRS is not a document life cycle management system; there are many commercial products thataddress those issues. The internal procedures involved in creating and reviewing documents arethe focus of such systems. Some of these systems offer interesting and useful functionality, suchas remote collaborative editing and annotating. These systems do not necessarily compete withthe components of NTRS because NTRS is focused only on the customer side of searching andretrieval. While NTRS could be expanded to provide access to different document types, only"finished products" are currently indexed in NTRS. NTRS provides access only to publicationsthat have passed through the existing approval and review mechanisms; it does not directlyaddress internal approval processes prior to publication.Initially, attempts were not made to convert existing paper documents to an electronic form,although it is desirable that all significant publications such as NACA reports eventually beaccessible via the WWW. The more difficult problem of adding electronic access to legacycollections is the focus of other projects (Smith, 1992). The results of these projects will beincluded in NTRS when they are available.Services ProvidedThe NTRS provides the following services to the user:1. A single standard interface to multiple NASA technical paper databases.The NTRS is a virtual "wrapper" script for several different NASA-wide Internet databaseservers. It simplifies the user interface for all servers to a single, common user format. The NTRSscript hides the operational differences of each of the servers from the user and submits theproperly formatted query to each participating report server. This is a widely distributed databasesystem, allowing each site participating in NTRS to update and maintain data locally, thuseliminating the need for central administration of the system. Figure 2 shows the interface ofNTRS to the various servers.2. Simple and rapid searches for information on NASA technical publications.Since the NTRS is available via the WWW, the user interface provided is a common, familiar,easy to use "point-and-click" style. WWW interfaces are available for most UNIX workstations,Macintosh and Windows-based PCs. The information is available to users' desktops 24 hours a

day. Currently, searching is limited to only the abstract and bibliographic information, not the fulltext of the document. At this time, only Astrophysics Data System (ADS) database implementsfielded searches. All other databases do not allow restricting search terms to specified fields(author, title, abstract, etc.).3. Rapid delivery of complete copies of technical reports.If a report is available on the NTRS system, the user may choose to download a copy with asingle click. The reports are currently stored in either PostScript or Hyper Text Markup Language(HTML) format. This is a significant savings in user time and effort. Retrieval time is measured inseconds, not days. If a report is not available on-line, information is available to allow the user toorder a paper copy via traditional means.NTRS Architecture and HistoryMuch of the design and toolset for implementing NTRS was developed for the Langley TechnicalReport Server (LTRS) (Nelson, 1994a & 1994b). LTRS began in 1993 as an anonymous filetransfer protocol (FTP) server with over 100 LaRC formal technical reports. While anonymousFTP was fine for those familiar with computers, it intimidated most casual users. The transition ofLTRS to a WWW interface integrated keyword searching and document retrieval, and allowedcasual computer users to access the publications.When the LTRS proved useful for researchers at LaRC and beyond, it was desirable to havesimilar access to non-LaRC NASA publications. Specifically, the implementors of NTRS desiredeasy electronic access to publications at other NASA facilities. The shell scripts, methods, andother products developed for LTRS were shared, so that LTRS-like nodes could be implementedat other sites.Thus, each participating facility was able to initiate and maintain its own technical report server,and all that was needed was a way to provide a level of integrated access. The NTRS home pagewas developed as a common gateway interface (CGI) perl script that presents users with a singlepage from which to perform a unified search. While the individual centers maintain their ownpublications servers for administrative efficiency, the NTRS page provides users with a single,integrated search facility.Overview of Related WorkTechnical publication servers have developed in a number of scientific communities, most notablyphysics (Ginsparg, 1994), computer science (Davis, 1994) & (Maly, 1994), and astronomy (VanSteenberg, 1992) &

(Accomazzi, 1995).These specific scientific communities led the way in electronic documentexchange, partly because the communities are heavily computer oriented and the majority haveaccess to the Internet. The aerospace community is becoming more Internet-capable with time. Inaddition to servicing the primary aerospace customers, NTRS provides a well-known location forsecondary customers to gain access to NASA research activities for potential technology transfer(Nelson, 1994c).NTRS provides a different service than common CD-ROM products. While there are many usefulCD-ROM products available, they suffer the same distribution problems of hardcopy. Unless theCD-ROM is available on the Internet, the fact that one site has a given CD-ROM does nothing toassist a site without the CD-ROM. In game theory terms, CD-ROM distribution is still a zero-sumgame. NTRS is a non-zero-sum game in that although only one copy exists, it is accessible toeveryone simultaneously.NTRS is different from previous NASA "electronic library systems" such as NELS (Smith, 1993)and NAM (Hunter, 1993) in several areas. The most obvious is that NTRS is provided in thecontext of the World Wide Web. NTRS does not introduce a custom client program withspecialized features. The other difference is that NTRS is limited to the domain of technicalpublications, and does not directly address non-publication information and data. While themodels developed for NTRS are applicable for a variety of uses, the current scope is limited totechnical publications.Software Implementation DecisionsQuick access to technical information is helpful to researchers. Most technical work is built uponprevious work, so it is desirable to have both historical and current work easily available fromwithin the same information system, frequently referred to as "hypertext" systems. Hypertextsystems allow information objects to be "linked" to each other in an intuitive manner. The mostsuccessful hypertext implementation to date is the World Wide Web. The Hyper Text TransferProtocol (http) used in the WWW is efficient in enabling a distributed knowledge network, andthe uniquely extensible Mosaic offers an elegant interface.The Wide Area Information Server (WAIS) (Kahle, 1992) software was chosen to implementsearching for the components of NTRS for several reasons. First, WAIS is freely available.Second, WAIS is a simple, generic search engine that is supported by many different platformsand integrates into a WWW environment easily. The source code is available, allowing sitecustomization where necessary. While WAIS is an attractive implementation for the NTRSsearching requirement, NTRS does not directly depend on WAIS. Should a successor to WAISappear, it would not be difficult to replace WAIS transparently with another search engine, andthe user should notice little, if any, difference. The citations for the publications in the variousservers are archived in a format different from their presentation. For example, most of thecitations are stored in "refer" format (Lesk, 1978), even though the user only sees a properlyformatted HTML citation (Fig. 3). Separating the archival format from the presentation formatwill allow for easier transition to other successor systems and the sharing of data with other nonHTML or even non-WWW systems. Refer format was chosen because it is simple and it is easy to

write various translators for refer - HTML, etc. There is no reason a richer archival formatcannot be used, such as bibtex (Lamport, 1986), RFC-1357 (Cohen, 1992) or even MARC.Tools and Customizations Resulting from NTRSA number of