Art Libraries Society of North America 32^nd Annual Conference
Roosevelt Hotel, New York City, New York, April 15-21, 2004

Session XVI

Digitizing Columbia

Panelists:

Robert Carlucci, Ph.D., Director, Visual Media Center, Department of Art History and Archaeology, Columbia University

Jeremy Stynes, Educational Technologist, Visual Media Center, Department of Art History and Archaeology, Columbia University

Juliet Chou, Educational Technologist, Visual Media Center, Department of Art History and Archaeology, Columbia University

Peter Sommer, Ph.D., Director of Education for Projects and Services, Columbia Center for New Media Teaching and Learning, Columbia University

Roberta Blitz, Digital Collections/Art Research Librarian, Avery Architectural & Fine Arts Library, Columbia University

Moderator:

Barbara Sykes-Austin, Avery Architectural & Fine Arts Library, Columbia University

Recorder:

Deborah Sommer, Environmental Design Library, University of California, Berkeley

Session sponsor:

Casalini Libri

Moderator Barbara Sykes-Austin, from the Avery Architectural & Fine Arts Library, Columbia University, introduced the session and the panelists. Sykes-Austin noted that the panelists collectively represent the state of the art of evolving processes of higher education coupled with new technologies. She asked each panelist to take questions at the conclusion of his or her presentation.

Robert Carlucci opened with a brief description of the Visual Media Center (VMC) in the Department of Art History and Archaeology at Columbia University (http://www.mcah.columbia.edu/). The VMC provides support to the Columbia art history faculty for undergraduate and graduate courses. Carlucci reviewed the history of using digital media for instruction at Columbia. An early project was the Raphael Project, 1998-2001, a video program focused on Raphael’s frescoes of the Disputa and the School of Athens in the Vatican making extensive use of computer-generated animations techniques. Its disadvantages included labor-intensive, and expensive production, difficulty integrating with courses, and with outsourced animation work, slow. (The program is now available on DVD for easier integration in the classroom.) Around the same time, the Media Center began development of a 35,000 image departmental art and architecture database over a three year period to demonstrate to faculty the potential of basic digital image resource development. In a separate effort the Department of Art History and Archaeology developed a 35,000 image departmental image database over a three year period, to demonstrate to faculty the potential of continuous normal slide production.

Simultaneously the Media Center developed a QuickTime virtual reality (QTVR) resource, for example, panoramic movies that are especially effective for both interior and exterior architectural subjects, to support the development of pedagogical tools for interactive tours on the Web, allowing students to ‘walk through’ archaeological or architectural sites. Carlucci used Frank Lloyd Wright’s Fallingwater house in Bear Run, Pennsylvania, as an example of the 250 QTVR nodes currently available. See http://www.learn.columbia.edu/fallingwater/ to take the tour. Also mentioned was an interactive Web site called the Venice Project, which allows viewers to hover virtually above the city and to zoom down to the interiors of individual rooms, digitally reconstructed with original paintings replaced on the walls. This site is available only to the Columbia Art History faculty. In late 2002 the department expanded its clientele to include the some faculty members from Architecture and Anthropology, and formed the Visual Media Center (VMC) as the successor to the traditional Visual Resources Collection (slide library) with the goal to effect a complete transition to digital teaching.

At this point, Jeremy Stynes, whose responsibilities span the spectrum of Web site production including graphic design, site architecture, imaging, and coding, noted that the VMC seeks to answer the question “How can we create meaningful online experiences for the core curriculum?” The VMC’s Art Humanities Web Portal (http://www.mcah.columbia.edu/arthumanities/) is one answer to that question. The Art Humanities database is designed to support classroom discussions and museum visits and contains all of the images in the Art Humanities syllabus. During its first two years the database included color images online, as well as all course media from the Art Humanities syllabus; 1200 images were scanned from class slide kits. The Art Humanities Database can be searched by a variety of strategies coordinated with the syllabus and can be used to create customized and shareable portfolios of images. The Web site and Database contain 12 primary topics; each topic Web site is comprised of three sections: images, texts, and supplements. Using the Parthenon as an example, Stynes took the audience on a QuickTime virtual tour that was dazzling in its comprehensiveness and subtlety. The site includes images of architectural replicas, photos, and plans, as well as portfolio links drawing on images from the image database. Please see the Database Web site for a personal view: http://www.mcah.columbia.edu/arthumanities/html/arthum_parthenon.html.

Juliet Chou, who specializes in the combination of a wide variety of Web site design programs, especially Flash programming, to create new ways of exploring complex works of art and architecture, described the benefits of teaching Asian art using innovative technologies. Asian art is distinguished by distinctive physical formats that can be challenging to display for teaching or research. At best the nature of the objects being reproduced is distorted. For example, the scroll, because of its unbroken length, is awkward to reproduce in slides and books. Chou demonstrated a prototype project using the painted scroll “Mountain Villa” by Li Gonglin (or Kung-lin) (104?-1106). She showed an interactive page with three horizontal images containing Chinese characters, a cartouche, and an English translation of the poem which is part of the painting, respectively. In the case of this painting, there is a bibliographic problem of multiple copies. The demonstrated technology allows the viewer to reconstruct the original from multiple sources in dispersed geographic locations, as well as facilitating the comparative study of brush work. The Web site home page (http://www.learn.columbia.edu/mvilla/mvilla2.html) contains brief information; the images are available only to the Columbia University community. The future development of Asian Art digital teaching tools will be supported by a grant from the National Endowment for the Humanities, Division of Education Programs, and the programs developed under this grant will be freely accessible to the educational community on the Web.

Questions: Panelists provided additional information in response to questions: the CVMC uses VRA core metadata. Columbia University is testing ArtStor. A teaching module takes approximately five months of development; the necessary material can be gathered in two to three days. The modules are developed on demand.

Peter Sommer described the mission of the Columbia Center for New Media Teaching and Learning (CCNMTL) as the purposeful use of new media in teaching and learning. Funded from the Office of the Provost (Columbia's chief academic officer), the program is five years old and has a faculty focus. On July 1, 2003, the CCNMTL became part of Columbia’s learning services operation, under the direction of the Vice President for

Information Services and University Librarian, James Neal. The CCNMTL has worked with approximately 2000 faculty in 18 schools within Columbia University. Aided by ten educational technologists, the Center has supported work ranging from the construction of course web sites to 150 advanced projects in partnership with the faculty. Please see the CCNMTL web site for more information: http://www.ccnmtl.columbia.edu/.

Sommer outlined the Center’s design research approach. The first stage, beginning with the curricular context and identifying challenges and hypotheses, takes two hours to six months. The second stage includes designing solutions with educational technologies, the educational experience itself, and research and evaluation.

To underscore the importance of trying not to repeat past errors, Sommer showed an 1899 image which predicted the classroom of 2000. He noted that the classroom occupants are all white and all male; the faculty member dumps the lecture material into a machine which pumps the information directly into the heads of the pupils via helmets with tubes. Sommer suggested this image raises the question of whether the digital library is a machine or an institution, clearly opting for the latter alternative.

Sommer used the Mutimedia Study Environments (MSE) of Salman Rushdie’s novel Midnight’s Children as an example of CCNMTL’s work. The MSE was created to complement a stage adaptation which was at the Apollo Theater in New York City. Rather than merely advertise the stage production, the MSE was intended to bring it to the University and the public and to make it part of Columbia’s core curriculum. The site is image-rich, rather than image-centered. The viewer may read excerpts of both the novel and play script, which are enriched with embedded links to still and video images, brief dictionary or encyclopedia annotations, interviews with faculty specialists, Rushdie, and many others. The text remains on screen while viewing the secondary material. The tools used to construct the Rushdie MSE are available to faculty to create similar multimedia learning environments. Please see the Midnight’s Children MSE Web site (http://www.ccnmtl.columbia.edu/projects/mmt/mc/) for additional information. Access to the project itself is limited to the Columbia University community.

In conclusion, Sommer compared the idea of a culture of use which flourished in the 1980s when everyone worked separately in pre-network environment with a culture of infrastructure which characterized the 1990s during which lots of infrastructure was built and everyone was getting wired. In the emphasis on infrastructure the culture of use shrank. Sommer observed that in the 2000s we have begun to build back the culture of use once again.

Questions:

Is there a formal process for development with the faculty?

Answer: There is an online project proposal form. In some instances the proposal can be implemented with a single already available educational technology; in other cases solutions are invented for a particular proposal. The CCNMTL works on 25 to 28 projects at the same time. One third of the budget is externally funded; faculty do not need their own funding in order to submit a proposal. The tool used to create the MSEs is proprietary and still being tested; eventually the design tool will be made publicly-available.

Roberta Blitz presented a paper entitled Using Computational Linguistic Techniques to Harvest Image Descriptors. Columbia University Libraries’ Center for Research on Information Access (CRIA) received funding from the Mellon Foundation for a two year project called Computational Linguistics for Metadata Building (CLiMB) to investigate the use of automated techniques to create subject access for the retrieval of images. The project is led by Principal Investigator Dr. Judith Klavans.

Blitz outlined some of the problems for access to images. The traditional approach is labor-intensive and expensive. Project CLiMB addresses the question “Is it possible to harvest image descriptions for metadata?” CLiMB is employing computational linguistic tools to identify and extract proper nouns, terms and phrases from existing scholarly literature associated with image collections, for use as metadata. Using the example of the Pratt House by Greene and Greene, Blitz demonstrated how a collection could be made searchable at a finely-grained level of detail. See http://www1.cs.columbia.edu/~delson/cni/ for this example and others.

Blitz outlined the overall goals of the project:

· Research: Development of richer retrieval through increased numbers of descriptors

· Practice: Development of a suite of CLiMB tools

In essence, the project piggybacks on pre-existing scholarship, using scholars themselves as “catalogers” by utilizing scholarly publications.

Blitz described the process of extracting metadata as the Metadata Squeeze. Beginning with image collections and associated texts, the texts are scanned. Target object identification (TOI) (a unique, "canonical" name for a major art object) creates an “authority” list of terms to load into the CLiMB Toolkit. The list of terms varies from project to project. In the Greene and Greene prototype project names were of interest. In the North Carolina Museum prototype, the artist/creator and object title were important. Working with a museum collection catalog, an image is more directly associated with a descriptive text, in contrast to the Greene and Greene project, where the texts are more discursive.

The CLiMB Toolkit (http://www1.cs.columbia.edu/~delson/CLiMB/gui/dev/) is a software tool designed for use by image catalogers for the semi-automatic generation of subject metadata. Please see http://www.columbia.edu/cu/cria/climb/presentations.html for a link to Blitz’s PowerPoint presentation and http://www.columbia.edu/cu/cria/climb/ to view the CLiMB home page.

Questions: Blitz provided additional information in response to questions:

The project only takes out metadata terms, so there are no copyright issues. The project uses OCR and TEI scanning technologies.