next previous
Up: The NASA Astrophysics Data

8 Discussion and summary

As of this writing (12/1999), there are 524 304 references in the Astronomy database, 523 498 references in the Instrumentation database, 443 858 references in the Physics database, and 3467 references in the Preprint database, for a total of almost 1.5 million references in the system. Astronomers currently write approximately 18 000 journal articles annually, and possibly that many additional conference proceedings papers per year. More than half of the journal papers appear in peer-reviewed journals. These numbers are more than double what they were in 1975, in spite of an increase in the number of words per page in most of the major journals ([Abt 1995]), and an increase in number of pages per article ([Schulman et al. 1997]). At the current rate of publication, astronomers could be writing 25 000 journal papers per year by 2001 and an additional 20 000 conference proceedings papers. Figure 1 shows the total number of papers for each year in the Astronomy database since 1975, divided into refereed journal papers, non-refereed journal papers, and conferences (including reports and theses). There are three features worth noting. First, the increase in total references in 1980 is due to the inclusion of Helen Knudsen's Monthly Astronomy and Astrophysics Index, a rich source of data for both journals and conference proceedings which began coverage in late 1979 and continued until 1995. Second, the recent increase in conferences included in the Astronomy database (starting around 1996) is due to the inclusion of conference proceedings table of contents provided by collaborating librarians and typed in by our contractors. Last, the decrease in numbers for 1999 is due to coverage for that year not yet being complete in the ADS.

\par\end{figure} Figure 1: Histogram showing the number of refereed journal papers, non-refereed journal papers, and conferences (including reports and theses) for each year in the Astronomy database since 1975

The growth rate of the Instrumentation and Physics databases is difficult to estimate, primarily because we do not have datasets which are as complete as astronomy. In any case, the need for the organization and maintenance of this large volume of data is clearly important to every research astronomer. Fortunately, the ADS was designed to be able to handle this large quantity of data and to be able to grow with new kinds of data. New available item links have been added for new types of data as they became available (e.g. the links to complete book entries at the Library of Congress) and future datasets (e.g. from future space missions) should be able to be added in the same fashion.

As with any dataset of this magnitude, there is some fraction of references in the system which are incorrect. This is unavoidable given the large number of data sources, errors in indices and tables of contents as originally published, and human error. In addition, many authors do not give full attention to verifying all references in a paper, resulting in the introduction of errors in many places. In a systematic study of more than 1000 references contained in a single issue of the Astrophysical Journal, Abt (1992) found that more than 12% of those contained errors. This number should be significantly reduced with the integration of the ADS reference resolver in the electronic publishing process. However, any mistakes in the ADS can and will get propagated, so steps are being taken by us to maximize accuracy of our entries.

Locating and identifying correlations between multiple bibliographic codes which describe the same article is a time-consuming and sometimes subjective task as many pairs of bibliographic codes need to be verified by manually looking up papers in the library. We use the Abstract Service itself for gross matching of bibliographic codes, submitting a search with author and title, and considering any resulting matches with a score of 1.0 as a potential match. These matches are only potential matches which require verification since authors can submit the same paper to more than one publication source (e.g. BAAS and a refereed journal), and since errata published with the same title and author list will perfectly match the original paper.

When a volume or year is mismatched, it is usually obvious which of a pair of matched bibliographic codes is correct, but if a page number is off, the decision as to which code is correct cannot always be automated. We also need to consider matches with very high scores less than 1.0 since these are the matches where an author name may be incorrect. The correction of errors of this sort is ongoing work which is carried out as often as time and resources permit.

The evolution of the Internet and the World Wide Web, along with the explosion of astronomical services on the Web has enabled the ADS to provide access to our databases in an open and uniform environment. We have been able to hyperlink both to our own resources and to other on-line resources such as the journal bibliographies ([Boyce & Biemesderfer 1996]). As part of the international collaboration Urania (Universal Research Archive of Networked Information in Astronomy, [Boyce 1998]), the ADS enables a fully functioning distributed digital library of astronomical information which provides power and utility previously unavailable to the researcher.

Perhaps the largest factor which has contributed to the success of the ADS is the willing cooperation of the AAS, CDS, and all the journal publishers. The ADS has largely become the means for linking together smaller pieces of a bigger picture, making an elaborate digital library for astronomers a reality. We currently collaborate with over fifty groups in creating and maintaining cross-links among data centers. These additional collaborations with individuals and institutions worldwide allow us to provide many value-added features to the system such as object information, author email addresses, mail order forms for articles, citations, article scans, and more. A listing of these collaborations is provided in Table 6. Any omissions from this table are purely unintentional, as the ADS values all of our colleagues and the users benefit not only from the major collaborators but the minor ones as well, as these are often more difficult for users to learn about independently. Most of the abbreviations are listed in Tables 2, 3, and 4.


Table 6: Collaborators
Additional Collaborations Nature of the Collaboration
A.G. Davis Philip Scanning of Conference Proceedings
Academic Press (AP) Scanning of Icarus
American Astronomical Society (AAS) Citations, Scanning of AJ, ApJ, ApJL, ApJS, AASPB a, BAAS
American Institute of Physics Scanning of SvAL
Andre Heck Star Heads (Author Home Pages)
Annual Reviews, Inc. Scanning of ARA&A
Astronomical Data Center (ADC) D links to data
Astronomical Institute of Czechoslovakia Scanning of BAICz
Astronomical Institute of the Slovak Academy of Sciences Scanning of CoSka
Astronomical Society of Australia Scanning of PASA
Astronomical Society of India Scanning of BASI
Astronomical Society of Japan Scanning of PASJ
Astronomical Society of the Pacific (ASP) Scanning of PASP and Conference Proceedings
Astronomische Gesellschaft Scanning of RvMA
Astronomische Nachrichten Scanning of AN
Baltic Astronomy Scanning of BaltA
British Astronomical Association Scanning of JBAA
Cambridge University Press M links to order forms, Scanning
Central Bureau for Astronomical Telegrams (CBAT) Object searches
Chris Benn Astropersons.lis (Author Email)
EDP Sciences Scanning of A&AS
Elsevier Publishers E links to articles
General Catalogue of Photometric Data (GCPD) D links to data
Institute for Scientific Information (ISI) Citations
International Society for Optical Engineering (SPIE) M links to order forms
Korean Astronomical Society Scanning of JKAS
Kluwer Publishers M links to order forms, Scanning of SoPh
Library of Congress (LOC) Z39.50 interface, L links to data
Los Alamos National Laboratory (LANL) Preprint Archive
Lunar and Planetary Science Institute (LPI) Scanning, Object searches
Meteoritical Society Scanning of M&PS
NED N links to objects, Object searches
The Observatory Scanning
Royal Astronomical Society Scanning of MNRAS
SIMBAD S links to objects, D links to data, Object searches
Springer Verlag Scanning of A&A, ZA b
Universitad Nacional Autonoma de Mexico (UNAM) Scanning of RMxAA, RMxAC
University of Chicago Press (UCP) Reference Resolving

$\textstyle \parbox{12cm}{
$^{\mathrm{a}}$\space American Astronomical Society Photo Bulletin.\\
$^{\mathrm{b}}$\space Zeitschrift f\uml ur Astrophysik.\\

The successful coordination of data exchanges with each of our collaborators and the efforts which went into establishing them in the first place have been key to the success of the ADS. Establishing links to and from the journal publishers, changing these links due to revisions at publisher websites, and tracking and fixing broken links is all considered routine data maintenance for the system. Since it is necessary for us to maintain connectivity to external sites, routine checks of sample links are performed on a regular basis to verify that the links are still active.

Usage statistics for the Abstract Service (see OVERVIEW) indicate that astronomers and librarians at scientific institutions are eager to take advantage of the information that the ADS provides. The widespread acceptance of the ADS by the astronomical community is changing how astronomers do research, placing extensive bibliographic information at their fingertips. This enables researchers to increase their productivity and to improve the quality of their work.

A number of improvements to the data in the ADS are planned for the near future. As always, we will continue our efforts to increase the completeness of coverage, particularly for the data prior to 1975. We have collected most of the major journals back to the first issue for scanning and adding to the Astronomy database. In addition, we are scanning and OCR'ing table of contents for conference proceedings to improve our coverage in that area. We are currently OCR'ing full journal articles to provide full text searching and to improve the completeness of our reference and citation coverage. Finally, as the ADS becomes commonplace for all astronomers, valuable feedback from our users to inform us about missing papers, errors in the database, and suggested improvements to the system serve to guide the future of the ADS and to ensure that the ADS continues to evolve into a more valuable research tool for the scientific community.

 The other ADS Team members: Markus Demleitner, Elizabeth Bohlen, and Donna Thompson contribute much on a daily basis. Funding for this project has been provided by NASA under NASA Grant NCC5-189.

next previous
Up: The NASA Astrophysics Data

Copyright The European Southern Observatory (ESO)