About SPIRO:

Environmental Design Image Database of the
Architecture Visual Resources Library,
University of California, Berkeley

  1. Introduction
  2. Project Goals
  3. Database Design
  4. Web and Database Connectivity
  5. Online Cataloging
  6. Label Formatting & Printing
  7. Digitizing Images
  8. Digital Storage
  9. Public Access to SPIRO & Copyright
  10. Impact on Internal Procedures
  11. Access to AVRL & Its Collections
  12. Circulation of Slides
  13. Administration
  14. Bibliography
1. Introduction:
SPIRO is the visual online public access catalog to the 35mm slide collection of the Architecture Visual Resources Library (AVRL) at the University of California at Berkeley. The collection numbers over 250,000 slides and 20,000 photographs. SPIRO content is the responsibility of AVRL, while programming and system support is provided by UC Berkeley's Museum Informatics Project (MIP). SPIRO was named in honor of the late architectural historian Professor Emeritus Spiro Kostof, and stands for "Slide and Photograph Image Retrieval Online. SPIRO is freely available on the World Wide Web.

SPIRO permits access to the collection by seven access points which may be used independently or in combination:
  • historical period;
  • place;
  • personal name;
  • object name;
  • subject terms
  • source of image;
  • image identification number.

    Six of the seven fields have scrollable look-up tables. (The exception is the image id.) Use of the look-up tables are an integral part of any SPIRO search. SPIRO is not a keyword or natural language search engine. Because it is a database, patrons must enter the actual form of data in the correct box. The look-up tables display precise data values, or actual database content, to assist patrons use correct retrieval protocols. Digital images, or visual surrogates of the slides, help users identify the exact images for which they are searching.

    As of January 2004, SPIRO contained over 63,000 records linked to images, approximately 20% of AVRL's total slide collection. Thirty-three percent (33%) of the images in SPIRO come from images in books. These are produced in-house by copy stand photography under the fair use and educational copying provisions of the U.S. Copyright Law. Eleven percent (11%) of the images in SPIRO derive from copy stand photography from periodicals, also produced in-house. Thirty-eight percent (38%) of the images are donor- supplied, and eighteen percent (18%) are purchased from commercial slide vendors.

    Sybase is the relational database management system used for cataloging SPIRO records. Custom software named "slidedb", developed by UC Berkeley's Museum Informatics Project (MIP) permits online cataloging of the visual records into the Sybase database. Trained permanent staff catalog slides directly on-line. A long-anticipated Java version of SPIRO is on hold, pending campus planning.

    It is anticipated that, in the future, digital images from SPIRO will be projected directly into classrooms. For the present, however, SPIRO serves as the visual catalog to AVRL's collection. It is also a rich repository of digital images for PowerPoint presentations. Faculty and students on the UC Berkeley electronic domain are able to download large SPIRO jpeg images for use in PowerPoint presentations. Resizing and color correction of the large jpeg images is generally not necessary.

    2. Project Goals:
    University goals in partnering campus image and object collections with MIP were to permit online access to the diverse holdings of the University's numerous image and object collections. With online public access, images and objects can be made available to the entire academic community, permitting both access and preservation simultaneously. Access to and retrievability from complex, valuable, high use, and/or rare collections through multiple access points was another goal. A local AVRL goal was to facilitate faculty and student lecture preparation, including the provision of remote access from their homes and offices by providing visual representations of slides.

    Future desired improvements for SPIRO include
  • image sequencing for lectures, or soft-light box capability;
  • save list or portfolio creation;
  • direct searching from the look-up tables, eg, clicking on hot linked items rather than copying and pasting from look-up tables
  • perhaps automated barcoded circulation

    3. Database Design:
    AVRL's database structure was created by Librarian Maryly Snow after analyzing a variety of data structures and data types. In-depth experience with patron queries, original cataloging and subject indexing processes constituted the critical foundation for the structural and content design of the database. Additional data requirements were also identified. Both MARC and AACR2 were reviewed, so that the image database could contribute to the development of national image documentation standards. Another goal was to create an image database feasible and worthy of linking to a major bibliographic utility such as Melvyl or Gladis. Database support for online authority work and subject indexing were built into the database design from inception. Highest standards of image identification, description, and classification were sought to maintain par with traditional bibliographic protocols. Online cataloging and database design was approached with the belief that copystand photography images should be cataloged with the same documentation rigor paid to bibliographic materials and archival images.

    The result was a database structure of 195 data elements organized into 47 tables, vetted by John Ober. MIP programmer Steve Jacobson then implemented the tables in a relational database management system. Jacobson also developed the catalogers' database input program, "slidedb". Only minor modifications have been made to the program over the years. A hard copy of the database manual is available for short-term loan. Approximately forty of the fields are used in the course of normal online authority work, cataloging, and indexing.

    SPIRO's database is mappable to MARC (Machine Readable Cataloging), the Dublic Core, and conforms to the VRA Core 2.0, which recognizes work/entity and view cataloging levels. SPIRO conforms in part with VRA Core 3.0, which supports work/entity, view, and image source cataloging levels.

    The database was implemented originally in INGRES, and has since been exported to SYBASE. The text database is stored with other campus Sybase databases on a dedicated Sybase server, and is maintained by a full-time Sybase system administrator.

    4. Web and Database Connectivity
    SPIRO is not "live" on the Web. Web and database connectivity is achieved through a combination of Java servlets and cgi perl scripts written by MIP. Every evening the database tables are optimized to facilitate searching: SPIRO's numerous tables are denormalized with an extract produced of essentially three large, searchable tables. Thus, SPIRO is updated nightly. Look-up tables are updated monthly.

    5. Online Cataloging:
    The two AVRL catalogers, the librarian and library assistant, are responsible for online cataloging using "slidedb". Slidedb permits throughput of data into the Sybase database. Slidedb supports online authority control; subject indexing; call number verification and construction; label previewing, formatting, and label printing.

    The cataloging process can be delineated into three major steps. First, authority control work is entered for all proper names:
  • creator names;
  • building and object titles;
  • location names.

    All location names are linked hierarchically to their host location, eg, Berkeley to Alameda County, to California, to United States of America. Many objects are also linked hierarchically to their host objects, eg, Wurster Hall to the University of California, or the Lehman Pavilion to the Metropolitan Museum of Art. This hierarchic linking enables retrieval by multiple levels of specificity from the most specific to the broadest.

    Then, once authority work is completed and all necessary proper names have been entered, they are then linked together in a work, or heading, record. An work record must be created before a view, or slide, record can be constructed. Each 35mm slide represents one view of the work: thus, each work record may have multiple view records. The source of every slide is recorded, and is then linked to the view record.

    Subject indexing, is an integral and ongoing part of the cataloging process. Although the Art and Architecture Thesaurus (AAT) is available online through the Getty Trust, MIP has licensed a copy from the Getty Trust so that the AAT can be loaded directly into the catalogers' slidedb program.

    There are many advantages to an online resident vocabulary. These include:
  • efficient search and retrieval of terms by the catalogers;
  • internal spell check by the computer;
  • automatic term appends;
  • integration of local and AAT terms into one hierarchic display;
  • authority control of applied subject terms;
  • hierarchic searching capability for library users.

    AAT terms can be applied during the cataloging process at several different stages and in several places in the database. The three most important places in the database where AAT terms are appended are:

    1.Object record: the object title must be accompanied by an object type. AVRL appends the most specific object type term possible, preferring "solar houses" over "houses", and "abbey churches" over "churches", for example. AVRL has used over 700 different object type terms, most of which are taken directly from the AAT.

    2.The Work/Entity (Heading) Record: the work/entity record identifies a building or object by its creator, object name, and location, e.g. "Frank Lloyd Wright, Kaufmann House, Bear Run, Pennsylvania". Subject indexing terms that apply to the building or object as a whole are entered here. These work terms include materials, style, and period.

    3.The Slide or View Record: attached to the work record are one or more view records. Each online slide has at least one view record. The view record contains a verbal description of the visual content of the specific slide, eg, "plan, before remodeling"; "map, city growth". The view record also contains a range of creation dates for the building or object; and information to guide the computer in call number generation. The view record links to the source of the slide, eg. Progressive Architecture, January 1985, page 42. Subject terms that apply to the specific view can be appended at the time the view record is created. These terms refer to the particular contents of an image, rather than the work. Terms such as doors, windows, stairs, entrances, balconies are example of components of a work, rather than the work as a whole.

    AVRL enters AAT terms in their plural form, eg, subject = doors, or aerial views. The plural form is used to ensure future conformance with subject terms in bibligraphic systems.

    Because dates are not authoritatively ascribed in the field of architecture, they may vary according to author. Therefore AVRL does not store the date of object creation in the work/entity record, but in the view record, where the date may vary author by author. One could argue that this method is problematic: it is not the most efficient for the catalogers. Variable dates can confuse students and faculty alike, because one building may display a range of different dates. Nonetheless, this approach seems the most intellectually viable.

    All slides acquired by AVRL since 1992 are scanned and cataloged online in SPIRO. A limited program of retrospective conversion (recon, or conversion of existing slide records into the database) has been underway for the past seven years. As slides are selected from the collection for electronic student review, the slides are scanned and cataloged online. Thus SPIRO also supports electronic student review, or web-based study. By linking the database together with course web pages, AVRL can coordinate the creation and dissemination of scholarly information, and, in the process, achieve some efficiencies of scale and effort. See A History of Architecture and Urbanism course web site for course image review. Image review may be performed lecture by lecture as well as in random order in the "quizzer" for test preparation.

    6. Label Formatting and Printing:
    Slidedb software supports multiple label formats. The label type, selected by the cataloger, determines the order in which information is displayed on the label. AVRL uses six label types. For example, label type 1 organizes information by location, while label type 2 organizes information by architect name.

    35mm slide labels are printed on an Hewlett Packard LaserJet 4000 N printer on foil-backed laser labels. However, any printer that feeds labels straight through the machine can facilitate the proper placement of the six point type on the tiny labels. After the labels are affixed to the Gepe slide mounts, the labels are overglued with Elmer's Glue or overtaped with book tip-in tape (J-Lar clear polypropylene tape with permanent acrylic adhesive) to protect the print from dirt and grime; to promote long-term adherence of the paper label to the plastic mount; and to improve readability and longevity.

    7. Digitizing:
    Most slides are scanned in-house by student employees on a Nikon LS-4000 slide scanner with its proprietary color management system turned off to increase scanning time. Earlier slide scanners we have used include the Nikon LS-2000 (1999-2001), the Nikon LS-1000 (1997-1999), a Kodak Rapid Film Scanner 2035+ (1993-1997), and a Barneyscan (1988-1992).

    AVRL scans at 24 bits per image (8 bits each for red, green, blue, or RGB). AVRL archives all images as raw tiff files, approximately 56-60 megabytes each uncompressed, 3600 x 5500 pixels on DVD-Rams and CD-Rs, which are burned in-house. In addition, the raw tiff files are also archived by MIP on a tape robot. All scans are color-corrected and batch processed using DeBabelizer software to convert images into two jpeg sizes: a thumbnail 128 pixels on the longest side, and a larger jpeg of 650 pixels on the longest side, 96 pixels per inch. The large jpegs are suitably sized for PowerPoint presentations.

    Slides acquired from commercial slide vendors are licensed. Digital scans are purchased or scanned in-house with written permission of the vendor. Slides acquired from donors are scanned with verbal or written permission of the donor. Image license agreements have been negotiated with the assistance of UC Berkeley Business Contracts Office and Information System and Technology’s Software Licensing Office.

    8. Digital Storage:
    AVRL's jpeg files are transferred across the campus network to their host Unix server, supported by Information System's and Technology's (IS&T) Central Computing Services. They are stored compressed, and display when invoked as uncompressed.

    9. Public Access To SPIRO and Copyright:
    Public access to thumbnail images in SPIRO is provided as a public service by providing architectural and art documentation and image availability information. SPIRO is not password-protected and is available as a shared resource for educational, research, scholarly, and cultural institutions in order to encourage the development and dissemination of cultural documentation. The majority of images in SPIRO are copyrighted. Slides have been acquired through a variety of means: through copy photography by the Architecture Visual Resources Library under Sections 107 Fair Use and Sections 108 Library Copying of Title 17 U.S. Code; through purchase from slide vendors; through solicited donations. AVRL negotiates digital permissions from all vendors and donors for digital display on SPIRO.

    Faculty, staff, and students of the University of California, Berkeley see two image file sizes: a small thumbnail and a larger jpeg. The general public, those not located on UC Berkeley's electronic domain, see only low resolution thumbnail images, in accordance with license agreements. The thumbnail images serve two purposes: locally they are a finding aid to the actual 35mm slide in the Visual Resources Library. Remotely, they serve as a visual reference to the location of the original image in a book or periodical, or to its availability from a commercial slide vendor, or from AVRL's collection of donated and licensible image rights. All images are linked to their host item, either by author-title-imprint and page for a book, title-month-year for a periodical, or slide vendor name-address-telephone-slide number for a commercial slide. Many slides with donor sources are available for licensing for publication. Hot links to vendor web pages and email are provided when available. A link from SPIRO directs users to a statement on copyright issues and best practice.

    10. Impact on Internal Procedures
    AVRL's internal slide acquisition and production procedures have been heavily impacted by the development of SPIRO. Procedures are frequently streamlined and revised as changes in technology warrant. Accessioning slides into AVRL's circulating collection takes much longer when supporting a digital online catalog: scanning, color correcting, image archiving, documentation authority work, subject indexing, and online cataloging have been added to what was once a fairly simple process. Online authority work and call number verification is time-consuming and complex. Making a system transparent to a variety of users is demanding.

    However, the increase in time and complexity of procedure has resulted in tremendous benefits, both locally and remotely. Prior to the advent of SPIRO, access to slides was by one access point only. Library patrons without a knowledge of architectural history and those who could not remember who designed a particular building had a difficult time retrieving slides. Patrons searching for subjects such as building materials or building components such as stairs and belvederes were especially disadvantaged by the lack of subject access. The multiple access points provided by SPIRO permit users, whether local or not, whether knowledgeable or not, to retrieve slide images regardless of how they have been classified or where they have been filed. SPIRO also points users to the source of particular images, serving as a visual index to thousands of books on architecture. The so-called institutional memory of the visual resources curator has been codified and made accessible, and access to materials has been increased exponentially. Digital images are available for both classroom teaching and student review. Quality control measures have been implemented to ensure that digital images are correctly linked to their catalog records, that cataloging and subject indexing is appropriate. Computer control of data reduces typographic errors on labels.

    11. Access to AVRL and Its Collections:
    AVRL's 35mm slide collection circulates locally to UC Berkeley faculty and graduate students. All new users of the collection are required to attend an orientation. Duplication of slides is not permitted. Loan periods range from 24 hours to 7 days. Faculty teaching in the Architecture Department may borrow slides for up to 7 days. All other library users may borrow slides for 24 hours. AVRL is not open to undergraduate students, except to borrow slides of their own coursework.

    Instructors from local educational institutions and local design professionals may borrow slides from AVRL for up to 24 hourse for a fee. Click on Outside Borrowers for further information on fees. AVRL, like most visual resources collections, does not participate in Interlibrary Loan.

    12.Circulation of Slides:
    Automated barcode circulation has been a dream since SPIRO's inception in 1985, but has not been implemented. Library patrons continue to sign out slides on a photocopy machine equipped with a 500 watt photoflood back light. The backlight illuminates the slide images on the photocopies.

    13. Administration:
    The Architecture Visual Resources Library has been managed by Librarian Maryly Snow since 1979. She has been involved with the development of SPIRO since its inception in 1985. Librarian Snow reports to the Chair of the Department of Architecture in the College of Environmental Design. Library Assistant Jason Miller is the major cataloger and circulation manager. Senior Photographer Steven Brooks supervises 35mm slide, photograph, and digital image production and archiving. The Architecture Visual Resources Library has a total staff of 3.50 FTE (full-time equivalent): 1.0 Librarian; 0.75 FTE Library Assistant IV; 0.75 Photographer 10 months (equivalent to 0.625 FTE); 1.5 FTE work-study student employees per week during the academic semesters. SPIRO's database and web connectivity is supported by MIP programmer Susan Stone.

    AVRL occupies over 1,800 square feet of space, adds approximately 6,000 slides and digital images per year to its collection, produces study photographs for student review of instructional materials, and circulates its slides to faculty and graduate students of the University of California, Berkeley.

    The Architecture Visual Resources Library is funded and administered by the Department of Architecture, University of California at Berkeley.

    14. Bibliography On SPIRO:
    Note: All articles are by Maryly Snow, unless otherwise noted. References are shown with most recent titles first.

    "SPIRO, the Visual OPAC of Berkeley's Architecture Visual Resources Library", in Design Net: Knowledge e Information Management per il Design edited by by Paolo Ciuccareilli and Perla Innocenti. Milan: Edizioni POLI.design, 2002. pp. 87-96.

    "Automatic, Required, and Optional Subject Indexing of Architectural Images", in Visual Resources Association (VRA) Bulletin, vol. 24:4, Winter, 2000

    "Millennial Musings on Image Databases: Prerequisites for Planning and Evaluation", in Art Documentation, vol. 17:2, Fall 1998, pp. 17-23

    "Digital Images and Fair Use Web Sites", VRA Bulletin, Winter 1997, vol. 24:4, p. 40-43. Available on the Web at UT Austin Office of General Counsel

    "Fair Use and Licensing Agreements: Digital Permissions in the Slide Library", in, VRA Bulletin, vol. 23, no. 2, Summer 1996.

    " SPIRO: FAQ - Frequently Asked Questions About SPIRO", in, VRA Bulletin, vol. 21, no. 2, Summer 1994, pp. 13-18.

    " SPIRO and ImageQuery at the University of California at Berkeley", in, CHArt (Computers and the History of Art), Vol. 4, Part II, June 1994, pp. 43-53. (ImageQuery was UC Berkeley's proprietary networked image database software in use before the advent of the World Wide Web.)

    "Database Design Issues for a Non-MARC Visual Online Public Access Catalog", in, Information Technology, It's ForEveryone!: Proceedings of the LITA Third National Conference, Library and Information Technology Association, Denver, September 13-16, 1992, pp. 182-190.

    "The AAT Browser at the University of California, Berkeley" (has images of ImageQuery SPIRO), in, AAT Bulletin, no. 19, 1991, pp. 10-14. (This article has images of SPIRO displayed on ImageQuery software. This browser has been replaced and is no longer in use)

    "Access to Diverse Collections in University Settings: The Berkeley Dilemma", in, Beyond the Book: Extending MARCfor Subject Access, G. K. Hall, Boston, 1990, pp. 203-224. Co-authored with Howard Besser.

    "Visual Depictions and the Use of MARC: A View from the Trenches of Slide Librarianship", in, Beyond the Book: Extending MARC for Subject Access, G. K. Hall, Boston, 1990, pp. 225-235.

    Revised January 9, 2004