Multimedia, hypertext and information access (MHIA) covers some of the most technologically important areas in the broad field of information technology, computing and communications. Enormous investment has been made there in the last decade to develop faster networking, tailored computer systems, usable authoring software, integrated digital libraries, edutainment packages and a wide variety of applications. Low cost storage and ubiquitous networking have moved electronic publishing, hypertext access, search methods, and multimedia access into the main stream of the rapidly growing World-Wide Web. Tens of billions of dollars of our economy relate to these technologies that now underly the Information Age and flow over the Information Highway. Some of the most innovative research work in the science/engineering world deals with solving hard problems in compression, processor design, computer interfaces, and communication with acceptable quality of service.
Carnegie Mellon University (CMU), Cornell University, University of California at Berkeley (UCB), and Virginia Tech (VT) have played a lead role in many advanced research efforts in this field. All have been involved in large projects in the digital libraries area; CMU and UCB are part of the NSF/ARPA/NASA Digital Library Initiative. Cornell and VT play key roles in the Networked Computer Science Technical Report Library (NCSTRL) that will be an important part of the proposed project, and the other two were involved in the ARPA CSTR project that was a precursor. CMU, UCB and VT were all among the very first universities to work with digital video in 1989. Cornell, UCB and VT have played leading roles in the area of information storage and retrieval. UCB is at the hub of the information provision activities of the UC system (with the Office of the President nearby), and so is uniquely positioned to connect with that larger group of institutions. For example, UCB has developed and distributed software for multimedia networking and digital video (MPEG) handling. CMU has been home to extensive work leading to commercial products in hypertext (e.g., the KMS system) and information retrieval (e.g., CLARIT, Lycos). All of these institutions have been integrating their research into educational contexts but will do so to a much greater degree as a result of this project.
Surprisingly, there are no generally approved curricula in the MHIA area! Further, few institutions have any regular courses in any of the subfields!
Hence, curriculum will be developed for MHIA in accordance with the two leading computer professional societies: ACM and IEEE Computer Society. In particular, this effort has been endorsed by the ACM Special Interest Groups on hypertext (SIGLINK), information retrieval (SIGIR), and multimedia (SIGMM) as well as the IEEE-CS Multimedia Computing Technical Committee. We will help develop a unified curriculum during the term of this project, to be co-published by all of these groups and widely disseminated. We also will work with other engineering groups around the nation that are interested in curriculum to help them meet the rapidly growing need for trained professionals in the MHIA area.
Meetings at annual conferences (e.g., the Courseware, Training and Curriculum in Information Retrieval workshop held in August at SIGIR'96, organized by PI Fox) will lead to recommendations for curricula in each sub-area of MHIA. Also, each year, on a rotating basis in connection with these workshops, there will be an integrative meeting to address the overall MHIA curriculum --- to ensure that it is suitably balanced and meets the general needs of the research, development and application communities. We believe that this open arrangement is likely to lead quickly to concensus and to reflect the needs of all interested segments --- computer science and engineering, as well as universities, government, and industry.
The philosophy for curriculum is to support a number of approaches to pedagogy, especially those that actively involve students, and to make recommendations for popular options. At the foundation will be a large number (200-500) of knowledge modules and supporting resources. The curriculum will suggest which and how these can be used for:
Each interested institution will identify the best approach to introducing MHIA in programs like computer engineering, computer science, electrical engineering, engineering science, industrial and systems engineering, senior engineering project courses, etc. Thus, current programs can evolve at an appropriate pace to cover more and more of MHIA, or can jump right in by adding whole courses or even sequences.
This project can begin with a rich base of courseware already developed at the four participating universities. See for example http://ei.cs.vt.edu/courses.html which is the root for a hyperbase of over 4500 WWW pages. The courseware will be integrated into a comprehensive, distributed digital library of resource materials:
By its very nature, the MHIA field is ripe for such materials. This is one of the rare fields where student projects can readily turn into courseware, and where application of new research results (e.g., devising authoring tools for multimedia training) can directly lead to improvements in learning. Thus, new results from research labs are amenable for classroom presentation or exercises (e.g., IBM's Query by Image Content demo on the WWW is used in a Virginia Tech multimedia course exercise). Good success had been obtained at Virginia Tech in applying this approach, leading to a senior class on MHIA, an introductory graduate course on Information Storage and Retrieval, and a senior honors class on multimedia.
New materials will be added to the digital library as part of the normal efforts of the professional societies --- a novel approach to coupling associations with education. All conference proceedings (from IEEE-CS events and those of ACM and its relevant SIGs) and journals of the societies may be made available electronically to project participants, with suitable intellectual property rights management policies enforced. All presenters of demonstrations, videotapes, tutorials and other conference-related activities will be asked to contribute their materials to the emerging repository. Submission specifications will be given to authors upon acceptance of their works, and project staff at VT will check and catalog each entry. The PIs will rate submissions and attach recommendations regarding educational utility, ensuring easy incorporation into curriculum.
The four lead universities all will have high speed Internet connections (DS3 or higher) that will enable videoconferencing to support periodic person-to-person or co-PI project meetings. That will unify the project team, which will be further brought together by joint course offerings between the four sites, and through MBONE and other methods, to other locations. UCB has already demonstrated the feasibility of this approach by offering multimedia courses through the MBONE.
Finally, the PIs will work with industrial partners to assist with coop programs in the MHIA field. Students at coop sites will be encouraged to access university courseware repositories and modules in the digital library whenever they find that they lack sufficient background for assigned work. This process can be extended to a lifelong learning support framework, and that concept will be explored during the course of the project.
The four universities involved will collaborate to develop curriculum for MHIA, under the aegis of ACM and IEEE-CS. VT will play the lead role, since with NSF CISE Education Infrastructure and Research Infrastructure grants awarded in 1993 it has begun courseware development and developed a good infrastructure base, as well as collected materials for a digital library in computer science. PI Fox directs the U.S. Department of Education funded National Digital Library of Theses and Dissertations and will ensure that MHIA-related theses and dissertations from that growing collection will be linked with this project. Also, Fox serves on the ACM SIGIR Education Committee (and was SIG chair 1991-95) and as co-chair of the SIGMM Education Committee.
PI Stevens of CMU has led work on the IEEE-CS Multimedia Computing TC, as well as on CMU's digital library work involving multimedia resources. He is part of the CMU committee developing an Entertainment Technologies Master's program. PI Rowe at UCB has applied MBONE and MPEG to disseminate instruction and video materials. He directs an NSF academic research infrastructure grant awarded in 1995 on Digital Media Storage and Transmission which will allow him to support this project with nation wide access to video storage. PI Smith has successfully engaged Cornell students in a number of multimedia courses during which they published high quality research papers.
These institutions cover the nation, from south to north and from east to west. They represent public and private institutions, of widely varying size. All are top-ranked research universities. Results of their research in MHIA will be made available through this grant.
Through connection with ACM and IEEE-CS, the main professional activities in this field will be harnessed to feed directly into education. Through connection with industrial partners (e.g., HP, IBM, Microsoft, US West), curriculum development will be guided to meet the needs of the communication and computing fields.
Because of the digital library base for all project deliverables, assessment, dissemination, and evaluation will be greatly facilitated. All usage of the digital library will be logged, and those adopting the results of the project will sign agreements requiring local logging and evaluation. The VT project team will analyze who uses what resources. Remote evaluation devices will be installed in all resources so that students' appeals to help systems or attempts to report problems will be recorded and analyzed in a relatively automatic fashion, so that large numbers of students can be serviced. Evaluations of resources, knowledge modules, courses, course sequences, and curricula will be collected and used to improve the digital library content, with assistance from contributing authors whenever that is possible.
Extensive evaluation work will be undertaken at VT. Because of an NSF CISE Research Infrastructure award (1993-98) on Interactive Accessibility, in which PI Fox serves as co-PI and Director of the Information Access Lab, there are six usability rooms along with powerful monitoring and logging software and hardware. New interactive materials added to the digital library will be tested by students in VT courses, and summaries of the resulting videotapes (showing user behavior and synchronized computer screens, along with a critical incident database) will be sent to developers of the materials to aid in their refinement. This in-depth evaluation will be supplemented with surveys as well as videotaped focus group discussions at the end of each semester. These will close the loop between research and education, leading to a much tighter inter-relationship.
Thus, our evaluation will be largely formative, which is appropriate for a project in which there is extensive development of materials and courseware. We will have summative evaluation through the surveys, focus groups and log-based analysis of actual use. In the case of two VT courses in the MHIA area which have been running for several years, we also will undertake comparative evaluations from year to year to measure the effectiveness of project-supplied enhancements.
The digital library will handle most of the dissemination of courseware. Curriculum will be released through society publications (e.g., CACM and IEEE Computer) and presented at annual meetings of all relevant societies. Feedback will be solicited so that continuous refinement of the materials can take place.
The digital library will function as a distributed system, initially involving the four lead universities, but expanding to the societies, industrial partners, and other groups as appropriate. Regional replication and other performance enhancing devices will be applied to ensure that dissemination will be possible to thousands of universities and tens of thousands of courses, involving hundreds of thousands of students.
As much as possible, this dissemination will be linked to the Networked Computer Science Technical Report Library (NCSTRL), wherein Cornell and VT play lead roles. That effort has support from NSF and ARPA and hopes to include reports from all leading computer science departments and research centers. The NCSTRL steering committee has expressed interest in expanding its role, and this project could help in that regard by way of broadening the types of materials included and the audience of users.
Each of the four institutions has extensive computing and networking infrastructure to support this initiative. By way of illustration consider the VT situation. In the Department of Computer Science there is an Information Access Lab, a Usability Methods Research Lab, and an Interaction Technology Lab, all part of the Interactive Accessibility project. All popular computer platforms are supported, as well as ethernet, FDDI, and ATM-based networking. Cameras, lighting, and a digital video capture, editing and post-production facility can produce broadcast quality videos, in digital as well as analog (e.g., BetacamSP, Hi-8, SVHS) formats. A state-of-the-art conferencing room is being finalized to support studies of CSCW, videoconferencing (with ATM support) and distance education. In addition, there are resources shared with the Computing Center that are fully available for this project, such as a $420K IBM SMP system with 4 processors, 512M RAM, 2.2 Tbytes hierarchical storage, and IBM digital library software. Around campus are about a dozen computer-integrated-classrooms with workstations for over 20 students, to support the type of lab-based courses that are likely to be most effective in the MHIA field. Full- scale deployment of ATM is underway, and VT is coordinating efforts for a state-wide WAN with DS3 and OC3 connections. Virginia Tech has more computers than telephones. Computer Science and Engineering students have been required to buy computers since 1985. VT is located in Blacksburg, VA, where about 40% of the population is connected to the Internet through the Blacksburg Electronic Village. By project completion, all dorms and thousands of apartments in Blacksburg will have ethernet access to the Internet.