BIBS: A Lecture Webcasting System

The Berkeley Multimedia Research Center (BMRC) at the University of California, Berkeley developed and operates the BIBS lecture webcasting system and, in cooperation with other campus researchers and organizations, conducted a series of experiments that attempted to assess the impact and usefulness of this technology. This report describes the design and implementation of the BIBS lecture webcasting system and the results of various evaluations of the system.

The BIBS system offers live remote viewing and on-demand replay of course lectures using streaming audio and video over the Internet. During the Fall 2000 semester 14 classes were webcast, including several large lower division classes, with a total enrollment of over 4,000 students and small upper division and graduate engineering courses.

The original design principles of the BIBS system were 1) that the technology must adapt to the teaching style of the instructor, 2) that the lecture webcasts are not intended to replace attendance at live lectures, 3) that operating the system must be cost-effective, and 4) that the system must be easy to install and use.

Components of the system are described, including:

• To managing video content, BMRC developed a database application and web pages for end-users to easily access the video residing on the system and for staff to manage the system.
• The BMRC Lecture Browser is a web-based player that presents to the viewer the speaker stream, presentation material, an index to the slide titles, and a keyword search interface. It is an important adjunct to the system as it allows for easy student access to video content.
• Servers and Software. The web server and DBMS system run on a Linux PC. This system serves all BMRC web requests, which averages over 400,000 page requests per month. The video gateway computers are Linux PC's.
• Lecture Capture and Video Gateways. Our approach to capturing lectures is to transmit the analog audio and video signals to a machine room in the building using either RF transmission on coax or fiber or a baseband signal on unshielded twisted-pair cabling (e.g., CAT5).

Costs

The cost of setting up and operating a system like BIBS is complex. The costs can be disaggregated into three components: 1) the streaming media, web, and database servers, 2) the audio/video equipment in classrooms, and 3) staff to maintain and operate the system. Each cost component is discussed in the following paragraphs.

• Streaming Media, Web, and Database Servers are run on different computers. A system built to service BIBS can run the web and database servers on one PC that costs approximately $5,000. A streaming media server that can handle at least 25 classes and 200 concurrent streams costs approximately $32,000. The system architecture for a permanent BIBS system should use redundant servers and network connections to improve reliability and availability. Redundant network connections are needed to improve access reliability in the event that a network interface or router fails.
• Classroom equipment. Each classroom, called a studio classroom, must be equipped with sophisticated equipment at a cost of between $25,000 and $100,000 for a limited webcasting facility, and between $250,000 and $500,000 for a sophisticated broadcast facility that has a control room with special-effects equipment and multiple projection screens for live two-way distance learning. Future development of classrooms at UC Berkeley should include a mixture of low-cost studio classrooms that can be used for webcasting and more expensive distance learning classrooms with two-way capabilities
• Staffing varies depending on the type of classroom used. The current cost of producing and webcasting lectures for a semester is $3,000-$4,000 per class, which includes OMS staff and a BMRC webop. We are experimenting with webcasts that have multiple video streams in order to replace expensive staff with intelligent software and user control.

Evaluation

Evaluations were conducted in Spring 2000 on student and faculty use of BIBS (e.g., student surveys, focus groups, usage statistics, and instructor interviews). A more in-depth evaluation of Chem 1A student and faculty use of BIBS was conducted in Fall 2000.

The primary use of the webcasts is to study for examinations. Students report they watch BIBS lectures because they did not understand material presented in lecture, because they wanted to review what the instructor said about selected topics, because they missed a lecture, and/or because they had difficulty understanding the speaker (e.g., non-native English speakers). We found that more than 50% of the students enrolled in some large classes view lectures and that as many as 75% of the lectures are played by members of the Berkeley community.

Results from these evaluations, in combination with other feedback received from students and faculty, suggest that lecture webcasting is a valuable service that enriches the UC Berkeley learning experience. Further study is required to accurately assess the pedagogical impact that lecture webcasts have on student learning.

Future Research and Development

Several directions for improving the BIBS system and further exploring the use of Internet technology in teaching and learning are discussed, including searchable text transcripts and support for the hearing impaired, authenticated access, distributed multimedia collaboration, extensions to the BMRC Lecture Browser, and campus commitment to audio/video infrastructure.

• Students found that the keyword search feature was particularly valuable when studying for examinations. It is likely that a combination of human editing and topic-specific training and filtering can produce better indexes.
• The current BIBS system does not limit access to the lecture archives. BIBS can be modified to allow various levels of authenticated access: registered students, the UC Berkeley community, and open public access.
• We advocate that a lecture webcast should include multiple video streams (e.g., speaker, presentation material, live experiments, views of other participants, etc.) and rich multimedia content. Further research should address this issue.
• BIBS, as it currently stands even with multiple video streams, is inappropriate for synchronous distance learning. BIBS can be a foundation for a distributed collaboration system but it will take considerable research and experimentation to produce an acceptable solution.
• Our philosophy has been to provide access at low cost. Much research remains to further reduce the classroom infrastructure operational cost. The objective should be fully-automated operation with user-control of the content being viewed.
• The BMRC Lecture Browser can be enhanced in numerous directions, including synchronizing student notes with a Lecture Browser title, a mechanism to reduce the time required to view a lecture, and creating short summaries of a video presentation.
• The video infrastructure and support at Berkeley makes it difficult to conduct experiments of the sort described here, and to build on those experiments. Just as the campus set an objective to provide a network connection in every classroom several years ago, we need to provide audio/video capability including cameras and microphones in every classroom.

Conclusion

It is time to move the system to a permanently funded service organization on the campus. While the current system has limitations for remote synchronous distance learning, specifically the inability to ask remote questions and the absence of a sense of presence for remote participants (i.e., a reverse video channel), several options exist for providing these capabilities and they should be explored. Continued research and development is required both on BIBS and other teaching and learning technologies.