Interview: Stephen Abrams
Stephen Abrams is newly appointed as digital library program manager in the Harvard University Library's Office for Information Systems (OIS), working on the Library Digital Initiative (LDI). Abrams joined the OIS team four years ago, as the LDI program was taking shape. Prior to his current OIS appointment, he served as development team leader in OIS. He has a bachelor's degree in mathematics from Boston University and he is currently working on a master's in art history at Harvard. He was interviewed for Library Notes on August 20.
LN
You're moving into a new leadership slot with the Library Digital Initiative. What's your portfolio?
SA
As the new digital library program manager, I'm charged with facilitating the ongoing progress of the Initiative.
This essentially has two components: one is internal to OIS and the other is University-wide. Within OIS, I'll be
participating in the development of appropriate technical architectures for new systems as well as
enhancements to existing systems. I coordinate a series of meetings in which we vet new development projects
with librarians and technical experts drawn from the entire University. Also, I chair the Digital Collection
Systems Steering Committee, which has direct oversight of the "orphan" infrastructure systems—those that don't
have a public face or active librarian or
curatorial constituency, such as DRS, NRS (persistent naming), Access Management, PDS (page turner), FTS
(text query), and TED (templated databases). I'll also be providing some longer-term thinking on issues of
strategic importance to Harvard's digital library program. Additionally, I represent Harvard's interests in various
national forums, such as the Digital Library Federation.
Since there is such an extremely high level of technical competence among my colleagues at OIS, my role in collaborating with librarians and collection curators outside of OIS is perhaps of greater importance. In cooperation with Wendy Gogel, digital library project liaison, and Sue Kriegsman, digital library project manager, I'll be providing advice and consultation to Harvard librarians as they develop and sustain projects funded through the LDI Challenge Grant Program.
LN
Until last January, MacKenzie Smith served in a capacity similar to your own. How will your responsibilities be
the same and how will they differ?
SA
My responsibilities are essentially the same. One of the particular strengths that MacKenzie brought to this job
was a vast network of personal relationships with librarians and curators all over the University. I can only
hope to try to approach the same level of personal interaction over time.
Unlike MacKenzie, I am not formally trained as a librarian. My background is in software development from the computer-science side. Luckily, within the collegial environment of OIS I've learned a great deal from my colleagues, who are extremely well-versed and articulate on all matters library-like. I prefer to look upon my not being a librarian as an asset, rather than a liability. Without pre-formed ideas of what is conventionally held to be not possible, I hope to be able to foster innovative thinking and new approaches to problems.
LN
You're not a librarian per se, but you have a great track record in information science.
SA
It's rather varied. As an undergraduate, I originally studied theatrical scenic design before switching hats
completely. I graduated from BU with a degree in mathematics. My undergraduate research generated large
numerical results that led me to explore—and later to develop—computer-generated graphics systems. I found
that I enjoyed the programming more than the math, so I continued to work in the graphics area along with
scientific visualization and user interface design, most recently as a research staff member at MIT. Much of my
initial work there was for the US Navy, but as Defense Department funding dried up, we began to work on
projects for other government agencies in ocean resource management. Those projects were essentially
information retrieval systems, which led me to the digital library world.
When the opportunity arose to specialize in that area here at Harvard, I took it eagerly, first as a senior programmer/analyst and later as development team leader in OIS. I already had some perspective on Harvard's libraries from the patron side as I'd been enrolled at Harvard in the graduate program in art history where I will soon—hopefully—be completing my thesis.
LN
What's your assessment of the Initiative after four solid years of effort? What have we accomplished and where
do we need to go?
SA
We've accomplished a phenomenal amount—and we started from scratch. The Digital Repository Service, or
DRS, is in place. It's a comprehensive system for the long-term managed storage of digital objects. These
digital objects are now discoverable through a variety of descriptive catalogs. These include the HOLLIS union
catalog, VIA [Visual Information Access] for cultural heritage images, OASIS [Online Archival Search
Information System] for archival finding aids, and HGL [Harvard Geospatial Library] for geospatial resources.
Behind the scenes, VIA and HGL both have sophisticated cataloging interfaces for entry of new materials.
Once discovered, relevant digital objects can be delivered to the user through a number of format-specific
mechanisms, such as the Image Delivery Service (IDS), for unitary images; the Page Delivery Service (PDS),
for structurally complex
page-oriented material requiring more sophisticated navigation; and the Streaming Delivery Service (SDS), for
streaming audio.
Underlying these visible componentsis a network of infrastructure "glue" that allows the entire system to function effectively and securely. This glue includes the Name Resolution Service (NRS), which provides digital objects with persistent, location-independent identifiers; the Access Management Service (AMS), which enforces access restrictions if necessary; and the Full-text Search Service, which provides comprehensive services for indexing and searching textual content.
We are working on new systems to facilitate the rapid implementation of searchable interfaces for data that can't be placed in existing catalogs due to policy restrictions or mismatches in topical scope. We're concerned with interoperability with courseware systems deployed by iCommons and FAS ICG. We are also investigating a new, high-level discovery service that will allow parallel global searches across all library systems, and enhanced, semantically-rich links between related digital objects stored in different systems.
When we began work on LDI, we took a bottom-up, systems orientation. This made sense because we were starting with nothing and were developing functionality incrementally. DRS, for example, was designed initially to handle and deliver single objects. Now, we need to deal with groups of related images—archival masters, production masters, deliverables, thumbnails, and so forth. So we added a mechanism for capturing inter-object relationships. Subsequently, we dealt with more complex structural relationships, such as those present in a digital volume composed of chapters, sections, and pages.
In LDI, we've reached a fairly mature stage. As a result, we need to adopt a more top-down, user orientation. Our users simply want easy access to digital stuff, and don't really care about what's going on under the hood to make that happen.
We need to concern ourselves with collection management issues. Today, a collection of heterogeneous content may be spread across the University in several repositories and in different descriptive catalogs. We need to ask—and answer—the question, "How can we help curators to deal with this content as a unified whole?"
LN
The University mandated and funded LDI to create the infrastructure needed to support our digital collections.
At some point will we consider the infrastructure complete?
SA
The infrastructure will always need to evolve in order to support new object types and data formats. Since
these have a relatively short useful lifespan, some level of continual development will be required. We have
spent a great deal of time designing our systems to be robust. But as the amount of content and usage increases
over time, we really don't know how performance and function will be affected. We are approaching collection
sizes and complexities where literally no one has any practical experience for us to draw upon. So we will have
to review, and if necessary, upgrade our hardware platforms and software periodically.
Another unknown facing us is digital preservation. We currently have in place the infrastructure for the persistent storage of digital objects. That is, we can always deliver the bit stream as originally deposited in our systems. But that's not the same thing as insuring that digital objects remain useful over time.
Digital content is fragile; when the tools necessary to create or display a particular format are no longer available, content stored in that format becomes essentially unusable. We need to plan on intervening proactively to keep valuable content alive. There is little experience or research that we can look to for advice in this area.
I think it's safe to say that the work initiated under LDI will never be complete.
LN
How widespread is the use of the digital infrastructure that we have?
SA
Outside of the HOLLIS catalog, the Harvard Libraries site [formerly known as the HOLLIS portal and
located at http://lib.harvard.edu] is probably the most utilized piece of digital infrastructure. It contains links to
about 4,400 electronic resources that are accessed some 200,000 times a month at peak periods. In our other
systems, we have over 300,000 digital objects stored in DRS, occupying
1.5 TB of storage. This includes 127,000 TIFF images, 106,000 OCR text files, 54,000 JPEG images, and
15,000 XML documents. Over 70,000 of these objects have been given persistent identifiers in NRS. VIA
provides access to 160,000 visual images, which are searched around 2,000 times per month. That is just
where we are today. A number of projects are underway to add significantly to our digital collections. For
example, the Fine Arts Library is beginning the process of adding 250,000 records and 100,000 digital images
to VIA. Of course, all of this activity pales in comparison to the HOLLIS catalog, which has nearly 9 million
catalog entries that were searched over 6 million times in 2001.
LN
Are we about to see a comprehensive effort to "populate" the infrastructure that is now in place?
SA
Yes. Through the LDI Challenge Grant Program, we are working with librarians and curators on a variety of
projects to increase our digital holdings in areas such as music, biomedical imagery, Latin American studies,
and botanical and archaeological source materials. Additional grant rounds will continue on in the same vein.
We are also investigating external funding for a large-scale effort at
providing digital access to significant portions of Harvard's enormous, and enormously valuable, collections.
Bringing these resources into the light of day and making them available to scholars really gets to the core
essence of the library's mission. It's tremendously exciting and satisfying to be involved in this effort.
Last modified on Tuesday, September 17, 2002.