background image
segments of the videos, rather than the
whole video.Therefore, we segmented the
videos manually based on human judge-
ment, for instance, by a story or event clip.
The duration of the videos in our collec-
tion ranges from several seconds to about
an hour (see Table 1 for details). In addi-
tion, we provide metadata at three levels:
entire video, segment and frame level.The
current metadata schema being used is
Dublin Core compliant, and we are an
Open Archives Initiative Data Provider.
with other digital video col-
lections such as Internet
Archive Moving Image
Collections (http://web movies/
movies.php).We also encour-
age anyone who owns videos
to make their collections
available from the video
archive. As of Spring 2003, the
archive provides about 2000
video segments (more than
0.5 terabytes of content), rep-
resenting 460 hours of video footage.
Among them are documentaries, educa-
tional videos, and ephemeral videos (most-
ly from the Internet Archive, http://www.
ince downloading a whole video clip is
time-consuming, video segmentation is
a fundamental step for the efficient access-
ing, retrieving and browsing of large digital
video libraries [Wang et al., 2000].We also
assume that most users want only short
he Open Video Project grew out of
experience designing the Baltimore
Learning Community Project
[Marchionini et al., 1997], which intended
to provide multimedia materials to middle
school science and social studies teachers.
Since 1999, the goals have evolved in three
directions: to create a publicly accessible
digital video repository for educators and
researchers, to develop and evaluate user
interfaces of digital video libraries, and to
develop open source digital library tools.
Currently, the collection contains about
2000 individual video segments, including
MPEG-1, MPEG-2, MPEG-4 and
QuickTime files.The videos were con-
tributed from various sources such as gov-
ernment agencies, universities, and personal
collections. In this article we first introduce
the current collection content and the
contribution sources, then describe our
user interface research issues, and finally
give a brief summary of the digital library
tools we have developed.
he first goal of the Open Video
Project is to create an open source
digital video archive for the research and
educational communities.To build this
open archive, we have acquired specific
video collections. In addition, we try to
instantiate and evaluate the Sharium con-
cept for digital libraries (DLs) [Marchio-
nini, 1999], which means inviting
contributions and gathering patron partici-
pation in DLs. Our videos were con-
tributed from various sources: US
government agencies such as the Library
of Congress and NASA, research projects
such as Carnegie Mellon's Informedia
Project (http://www.informedia.cs.cmu.
edu/), the University of Maryland's
Human-Computer Interaction Lab (HCIL:
and the Digital Himalaya Project (see Table
1), and also from personal collections such
as the Prelinger Archives (http://www.prel In addition, we collaborate
Figure 1 - Example of a storyboard surrogate
.org - 2003