Navy library hones search tool

Navy library hones search tool

NRL chief librarian Laurie E. Stackpole says Torpedo Ultra users get e-mail alerts about new journal content.

Service's researchers can browse texts of millions of scientific documents

By Patricia Daukantas

GCN Staff

The Naval Research Laboratory's Ruth H. Hooker Research Library recently launched Torpedo Ultra, a browser retrieval tool for the small institution's huge digital library.

What sets Torpedo Ultra apart from other digital library efforts is that it accesses the entire texts of many scientific journals stored on NRL hardware instead of on publishers' Web sites. The NRL library is rolling out state-of-the-art storage systems to handle the digitized text storehouse as it grows into the terabyte range.

Besides NRL scientists, researchers from four other institutions that belong to the National Research Library Alliance can use Torpedo Ultra. The member libraries are at the National Institute of Standards and Technology, NASA's Goddard Space Flight Center, the National Science Foundation and the Naval Postgraduate School. The Office of Naval Research also has access.

A library first

NRL started digitizing documents in the early 1990s and put its first version of Torpedo online for lab users in 1995, chief librarian Laurie E. Stackpole said. 'It was probably one of the first attempts by a library or anyone else to make the full content of scientific publications available at the desktop,' she said.

NRL's effort attracted the interest of several publishers, including the American Physical Society in College Park, Md. In 1993, NRL started digitizing several of the society's journals, although society officials wondered whether anyone would want the service. Torpedo's initial content consisted of 185,000 NRL technical reports'9 million pages' worth'and some society journals.

As Web technology began exploding, NRL decided to expand Torpedo's search capability. Stackpole said the NRL library ran a beta version of Torpedo Ultra last year before the rollout of the final version on Jan. 19.

Torpedo Ultra uses an off-the-shelf search engine, RetrievalWare from Excalibur Technologies Corp. of Vienna, Va. Progressive Technology Federal Systems Inc. of Bethesda, Md., did the systems integration and helped customize the software.

NRL's digital library now contains 220 journals published by Elsevier Science of New York; six American Physical Society journals; nine journals from the American Institute of Physics, also in College Park; 6,000 technical reports, mostly from NRL; 10,000 journal articles by NRL authors; and 2,000 NRL press releases. The library plans to license at least 175 other journals for digital access.

This year the Torpedo Ultra team will introduce a faster search interface and add cookies to track user preferences.

Researchers at member sites can subscribe to Torpedo Ultra's contents-to-go alerts, Stackpole said. The subscribers get e-mail messages with hyperlinked tables of contents when new journal issues are posted. The average user monitors about 20 journals by e-mail, she said.

Researchers at the member institutions benefit from having the documents mounted locally instead of retrieving them over the Web from publishers' sites, Stackpole said.

'We're able to ensure that the access to this information we have licensed is permanent,' she said.

R. James King, an NRL specialist in library information technology, said a researcher starts a Torpedo Ultra session by choosing the browse or the search option.

Torpedo Ultra can search for articles, press releases and technical reports together, King said. Concept-based searching, pattern matching and Boolean searching are available as advanced options. During beta-testing last year, Torpedo Ultra's developers found problems with its JavaScript interface on Apple Macintosh clients, King said. They changed the system to switch seamlessly to an alternate frames-based interface when it detects a Mac client.

NRL for some time has made the Science Citation Index'a huge bibliographic directory published by the Institute for Scientific Information in Philadelphia'available to lab users on CD-ROM.

Torpedo Ultra is being integrated with the Web of Science, an even larger database product from the institute.

Stackpole estimated that about 80 percent of the library's periodical collection, which covers physics, chemistry, materials science and electronics, is available electronically to lab scientists.

The rest consists of old issues that have not yet been digitized and journals to which NRL has no electronic subscription.

Because of copyright and licensing agreements with participating publishers, Torpedo Ultra is not publicly accessible.

Many journal publishers make their electronic content available only to paid subscribers or licensed sites.

More than half of Torpedo's content was originally in TIFF, Stackpole said.

For Torpedo Ultra, the library has switched to Adobe Portable Document Format, compatible with the widely used Adobe Acrobat plug-in reader.
Progressive input

The Adobe Acrobat approach bundles documents, said John R. Yokley, president of Progressive Technology. Previously, each page of an article had to be saved as a separate TIFF image. Under PDF, each document is one file, and its text is searchable.

The NRL library has a cooperative R&D agreement with the American Institute of Physics to digitize nine of the nonprofit publisher's journals, Stackpole said.

Torpedo Ultra searches are powered by an eight-way Sun Microsystems Enterprise 4000 midrange server with 7G of RAM, King said. The research library has a 500-disk CD-ROM jukebox for files that arrive on CD.

A newly acquired EV-5000 MegaDrive RAID system from DataDirect Networks Inc. of Chatsworth, Calif., has six 50G disks but can be upgraded to 2T or more, King said.

A 600G DataDirect EV-1000 RAID holds the PDF files, the Institute for Scientific Information database and the project's Web presence. A 324G Sun Sparc storage array is being phased out.

For regular backups, the library has an X2000 tape library from Exabyte Corp. of Boulder, Colo., equipped with Mammoth tape drives, King said. Each tape holds 30G of uncompressed data.

inside gcn

  • artificial intelligence (ktsdesign/Shutterstock.com)

    Machine learning with limited data

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group