Digital Asset Management—Off the Shelf or On Your Own

Page 2 of 3

Organization: Shoah Visual History Foundation
Assets: 52,000 video testimonies
DAM Solution: A homemade construction using tools from Advanced Digital Information Corporation, Sony, and Sybase.;;

Planning on training employees via the company intranet? Look at applications that offer Web-based communications. That way you don't limit user access based on platform (i.e., Mac, PC, or UNIX). Developing a Web site? Make sure you create a low-bandwidth and high-bandwidth version. Once again, you want to be inclusive of all online users.

But what if your company's mission is to educate the planet? You can't limit thinking to just the Internet because, in reality, only about 10% of the planet is online. Relying on Web-based tools alone will only make a small dent in your mission. Thus, you have to come up with a solution that doesn't solely rely on a computer or the Internet.

Such is the challenge for Survivors of the Shoah Visual History Foundation whose mission is to educate the world about prejudice, intolerance, and bigotry through videotaped testimonies of all Holocaust survivors and witnesses. Established by Steven Spielberg, the organization's goal is to videotape 50,000 survivors in 57 countries, preserve those interviews, and make the testimonies available to anyone.

Initial Issues
Prior to the Shoah Foundation, no one had ever attempted a project quite like this one. As such, there was no model on which the foundation could base its intellectual development so it had to start from scratch.

At its inception in 1994, the foundation hired Sam Gustman as its executive director of technology. Gustman came from the Army core of engineers where he worked in geographic information systems (GIS) managing large databases with maps. With the online video world at its nascent stages in 1994, "The GIS world was looked at as sort of the best place to bring over folks with the knowledge to index and manage that large amount of multimedia," says Gustman.

Producing content posed a monumental logistical challenge. To create a single video, the foundation had to identify a survivor, hire an interviewer, and a videographer, which required a worldwide scheduling system for interviews and tape shipments. "The hardest part during the collection was just the fact that we were in so many different countries," Gustman says. For foreign schedulers that didn't have any online access, computers had to be shipped and people who didn't know how to use a computer had to be trained.

Cataloguing and Indexing
Managing all this content rested on the shoulders of Gustman's IT team. "There weren't any systems out there that did everything from applying the keywords as we wanted to segmenting the video," he says. And as a result, his team spent seven years trying to piece together the best system.

Initially, they let the indexer who watched the video select keywords for indexing, as well as determine when a survivor's story began and ended. This method created two unforeseen problems: One was that during a survivor's interview, the first few minutes are spent mentioning names of people and keywords alone did not provide sufficient structure for people names. Second, determining the length of a story, usually about three minutes, was arbitrary, varied by who did the indexing, and increased the time it took to watch a video by a factor of 15.

To eliminate these indexing slowdowns, the first step was to conduct a pre-interview, or survey, by phone. The survey, about 40 pages long, contains the survivor's basic biographical information including all the names of people he or she can remember. This survey provides the top-level index card in the database to identify the survivor and video.

The second change was to eliminate the arbitrary time of story segments, which they decided to make one-minute long. They also chose to have indexers apply names and keywords from a thesaurus of 26,000 associated words. According to Gustman, by eliminating the need to determine the beginning and ending of a story, it cut down the indexers' review time from 15:1 to 2:1. Before Gustman finalized the new methodology in October 2001, the team had only indexed 4,800 video interviews. Since then, the foundation has indexed 16,000 interviews. At this rate, Gustman expects to finish cataloguing all 116,000 hours of interviews by 2006. The catalogue is managed through a Sybase database, and Advanced Digital Information Corporation's robotic tape system is used for both backup and near line storage.

Page 2 of 3