Creating a Virtual Classroom for Interactive Education on the Web

Dan Dwyer, Cornell Theory Center, 628 ETC Building, Ithaca, NY 14853-3801

dwyer@tc.cornell.edu, http://www.tc.cornell.edu/~dwyer/

Kathy Barbieri, Cornell Theory Center, 629 ETC Building, Ithaca, NY 14853-3801

barbieri@tc.cornell.edu

Helen M. Doerr, Cornell Theory Center, 526 ETC Building, Ithaca, NY 14853-3801

hdoerr@tc.cornell.edu

Abstract:: The World Wide Web provides new opportunities for distance education over the Internet. The Web, when combined with other network tools, can be used to create a virtual classroom to bring together a community of learners for interactive education. The Cornell Theory Center, a national center for high performance computing, is investigating the use of emerging network technologies for training computational scientists and researchers in the concepts of parallel processing. This effort is being built on electronic educational materials already on the Web and will evaluate the effectiveness of various collaborative tools.
Keywords:: Distance Education, Virtual Classroom, High Performance Computing, Computational Science

Introduction

The World Wide Web and other Internet-based collaborative tools have significantly enhanced the ability to train and educate electronically. Whether the materials are a stand-alone tutorial or a full-fledged on-line workshop, the Web provides significant new functionality in transmitting information to the student and providing forums for exchange. When integrated with tools such as listservs, Usenet newsgroups, annotation facilities, and video teleconferencing, the Web can greatly increase students' level of involvement in the training experience. The Web provides an effective mechanism for integrating many of these tools into a single interface and is an ideal tool for information which is itself rapidly changing. The Web is revolutionizing some areas of study through increased opportunities for learning and alternative formats for information.

The Cornell Theory Center (CTC) is a nationally funded high performance computing center founded in 1985 and located in Ithaca, New York. Our mission includes providing supercomputing resources and technical support to researchers at academic institutions and national laboratories across the United States. Educating researchers in effective and efficient use of the rapidly changing technology of high performance computing is a core component of our mission as a national center. Over 60% of the users of our facilities are located outside of Cornell University. Hence, use of an on-line information system and electronic tutorials has been central to our education and training activities from the beginning of the Center.

It is clear that the Internet, with its ability to connect people and information around the world, is already having a significant impact on education at all levels. The lofty goal of an interconnected global schoolhouse across remote corners of the world is getting closer. Virtual classrooms are being created today for educating scientists, engineers and mathematicians, both researchers and faculty, and will provide needed experience and basic knowledge for use with other student populations in the near future.

Education and Training at the CTC

The CTC provides educational opportunities for researchers, faculty, and students in the effective use of high performance computing for the exploration of scientific problems. Workshops for new and experienced users of the Center's resources are offered on a regular basis. Workshops include lectures on system features, optimization, introductory and advanced topics in scalable parallel computing, and visualization. Lectures are supplemented by guided hands-on laboratory sessions to ensure that participants gain needed experience. Over the past 9 years, the Theory Center staff has assisted hundreds of researchers in moving legacy serial codes to vectorized and now parallelized implementations.

On-line educational materials, which include lecture notes and lab exercises, are available through the World Wide Web. These materials are used by our staff during workshops and by others on the Internet who wish to provide instruction on these topics. In addition, the materials provide a handy access point for researchers and students to review concepts that were presented in workshops. Educational materials also include a set of tutorials especially designed for on-line learning. These tutorials are the recommended starting point for researchers and students desiring to become familiar with the use of the Theory Center's systems and scalable parallel computing.

In addition to direct education of users of our facilities, we use a "train-the-trainer" model to leverage our support effort. This is accomplished through our academic affiliates activities, known as the Smart Node Program. Each of the more than 90 participating universities sends a consultant to the Theory Center for a one week workshop each year. At this workshop, the consultants are updated on the latest technologies. They then return to their own campus to provide support to local researchers often by re-using our training materials in courses that they offer. Our on-line documents and training materials provide continuous access to materials for Smart Node consultants to use in training users at their local campuses. Extensive feedback from our Smart Node consultants over the past seven years has provided ample evidence that our materials are heavily re-used as is and frequently modified and incorporated into local materials to meet specialized needs. This was true even when we were in a 'paper' environment. With the availability of the Web, this re-use has increased dramatically.

Because of the rapidly changing technology in high performance computing, researchers who use our facility need to continually update their knowledge of state-of-the-art computational science techniques. As these users are spread across the United States, it is not feasible for frequent visits to the Theory Center to attend workshops. We have found that educating researchers in concepts and approaches to parallel processing is a staff-intensive effort. This spring we will continue to advance delivery techniques by creating a virtual classroom and pilot a course "Scalable Parallel Computing on the IBM SP2". This course will be offered electronically. It will draw heavily on existing on-line materials at the Theory Center and also link to other instructional materials available through the Web.

Electronic Education System

While our hardware and software vendors sometimes supply user documentation in electronic format, it is often "command" oriented, but our users often require "task" oriented materials (e.g., the user needs to learn how to run a batch job, without prior knowledge of the names of the five commands involved) and each vendor uses a different interface for access to on-line materials. Even now that computing vendors are creating World Wide Web servers, they are slow to include user documentation, but instead concentrate on marketing information. For all of these reasons we have made use of an electronic user documentation system since the founding of the center in 1985.

The Theory Center on-line information system contains educational materials in a variety of formats. We have developed more than 140 items specifically designed to be instructional tools that are referenced thousands of times each month. This electronic education system not only serves the CTC researchers, but also the international research audience, and provides materials that other educators can incorporate into their courses. Other supercomputing centers, such as the Maui High Performance Computing Center, Argonne National Laboratory, the Leibniz Computing Center and the Queensland Parallel Supercomputing Facility, maintain links to our materials.

What follows is a brief description of the types of educational materials in our on-line information system.

Tutorials: Electronic tutorials have been available for more than nine years. Tutorials are intended to be short, self-contained instructions on a specific topic (e.g., use of an optimization subroutine library). Each tutorial contains a short description, step-by-step instructions in using the utility, and an example code that can be executed. Researchers can read the tutorial and try the example, possibly taking the example code to use as a template for their own problem. These tutorials are task-oriented, not command-oriented.
Lab Exercises: Lab exercises have been provided in electronic format for the past two years. These exercises are intended to supplement lectures given at a workshop. The solutions to lab exercises are provided in a separate file to encourage the student to solve the problem instead of following the cookbook instructions found in a tutorial. Placing the lab exercises on-line has allowed workshop participants to more easily continue to work on the problems after returning home. Also, the on-line versions will be kept up-to-date where hardcopy lab exercises quickly become out-of-date.
Lecture Notes: During the fall of 1994 we began to maintain a complete set of lecture notes on-line. Formatted as a traditional "overhead foils", lecture notes are frequently a list of bullets which provide a guideline for the lecturer to use in presenting the materials. Lecture notes may not contain full details of the topic, but usually contain many references to other electronic materials for more information. In addition to providing continuous access to these materials for workshop attendees and other trainers, we have been experimenting with using NCSA Mosaic and Netscape as the mechanism for presenting the lecture. The students can follow the lecture on a terminal in front of them or can see the instructor's monitor displayed with a large screen projection system.

Future improvements to our system will be carried out to incorporate images, animation, and user-interaction forms into our documents. We hope to use the more advanced functionality provided by the Web and emerging search technologies to better integrate training materials, hardware and software documentation, reports on algorithm design and related research projects. Changes in these areas will continue the evolution to an increasingly effective on-line information system.

Plans for a Virtual Classroom

Distance education, in the form of a course offered across the Internet, is a logical next step for the Theory Center. This course is distinct from user-initiated, independent accesses to our on-line materials because it will have an instructor and a registered group of participants. This brings together a community of learners into a virtual classroom where they can interact with each other. Our goal is to create a virtual classroom, using methods that will fully engage the students in the learning process through an interactive, dynamic environment involving the student, on-line materials and an instructor.

Issues for Distance Education across the Internet

There are many topics that need to be explored before being able to offer complete workshops over the Internet. These range from technical considerations to sociological aspects of learning without the existence of a physical classroom. Much study has already been done in distance education. While distance education can provide efficient utilization of scarce resources, more fair distribution of knowledge, and timely dissemination of new knowledge, limitations in its effective use exist. Student motivation is critical. On-line instructors may need to take more time for the electronic course. Limits to the number of students may be required to maintain quality interaction. While generalizations in the effectiveness of distance education are dangerous, a common set of advantages and disadvantages are emerging (Rossman, 1992).

The Web solves some problems of traditional distance education and brings new issues to the forefront (Perron, 1994; Hurley, Marshall, McIntosh-Smith & Stephens, 1994). A leading effort in exploring this emerging world is the Global Network Academy (Butts, Reilly, Speh & Wang, 1994). The Theory Center will face a number of similar issues, but will also have some unique situations related to our field of instruction and the network-sophistication of our intended audience. Some of the issues addressed for our pilot course follow.

Target Audience: With traditional workshops, attendance is limited by the physical size of our training facility (about 50 students). Normally this means that attendees are users and potential users of our facilities -- researchers whose backgrounds, computing experiences, and expectations are somewhat similar. With an electronic course, the size of the class will be limited by other non-physical factors such as staffing requirements for account administration, limited consulting support, and CPU resources required for the class laboratory exercises. While we may now be able to directly impact a broader group of people, this group will likely be much more diverse in its background and computing experience. Will we want to set up individual computer accounts for this larger group on our system? Will we provide support for these people through our consulting phone and e-mail service? Should we instead look at a method to provide hands-on experience through a more generic method? Certainly there will be a number of security issues and the necessity for fire walls to protect our production computing environments if we wish to accommodate this larger audience.
Prerequisites: As we do for participation in our traditional workshops, we will require a basic knowledge of UNIX. With workshops held in our training facility we know what type of computer hardware, software and network connection our students will have. For an electronic course an Internet connection will be required, but will we require students to have a graphical browser for the Web? This requirement could exclude some students from participation, but would greatly increase the functionality that we can make use of for the materials. Even with a graphical browser, some students may find limited network bandwidth does not provide a satisfactory learning environment.
Learning Environment: Extensive feedback from attendees at our traditional workshops indicates there are two benefits to learning in a classroom setting which are independent of the material covered in the course. First, students find that the ability to get away from their other responsibilities and concentrate on learning the course materials greatly increases the amount that they learn and retain. Second, the interaction with other students and Theory Center staff, whether it is side discussions during the lab periods or simply listening to questions asked by others, often results in people rethinking their own computing techniques and suggest new solutions to their problems. An electronic course will need to have some of these same attributes in order to ensure that students find the experience worthwhile so that they complete the course. There needs to be some sense of an "instructor" for the students to interact with and the ability for the entire class to participate in questions and answers.

Audio-Visual Equipment in the Virtual Classroom

In a physical classroom there is a standard set of audio-visual equipment and tools available to the instructor. These might include a chalkboard, overhead projector, video cassette player, possibly a sound system, and even the textbook. Professional instructors know how to make best use of these tools. The virtual classroom will need equivalent equipment and tools in the form of network-based software applications. Some of these virtual tools have a relatively long history on the Internet and provide obvious applications. Other are still emerging and their potential use in a virtual classroom is not yet understood. With the appropriate design, the student should be able to take advantage of these tools without leaving the comfort of their favorite Web browser interface. Some of those considered for our classroom are;

The textbook: Many courses have a textbook used to guide the direction of the course or for background material. While our on-line materials provide some of this function, these may sometimes not provide all of the details that a users needs. In the area of high performance computing an example of a useful electronic textbook is the Computational Science Education Project (ORNL, 1994) sponsored by the U.S. Department of Energy. The growth of electronic publishing on the Internet should ensure a good supply of electronic textbooks over the coming years.
The chalkboard: Even when the lecture materials are prepared in advance and made available to the student, most instructors will make use of a chalkboard for further clarification of a point. In an electronic course, the instructor might make use of the shared whiteboard offered by a tool like NCSA Collage (NCSA, 1994) to answer a question from a student. Such tools allow images to be displayed, manipulated, annotated, and shared between two people or among a whole group.
Video cassette recorder: Use of mpeg movies and audio clips can be effective additions to textual materials. For our course on scalable parallel computing we might create a "cartoon" to illustrate how messages are passed between processors in a distributed system or record the animated output of a parallel trace tool. Audio clips of the instructor might be helpful in some instances to describe a particularly difficult point. Either can add some needed relief for what might otherwise be rather dry material -- just as an instructor in a traditional classroom might use video tape to keep the students' attention.
The sound system: An important part of any physical class setting is the personal interaction as questions are asked by the students. Allowing all students to "hear" the questions and answers helps everyone learn and encourages additional questions to clarify a point. Use of a listserv to redistribute e-mail questions or a usenet newsgroup are simple methods for sharing this interaction. A more dynamic question and answer period could be created using a chat session. For some types of courses a multi-user, text-based virtual reality, also known as a MUD (Multi-User Domain) or MOO (Multi-user domain Object Oriented), might be effective. An example of an experiment in this area is the Diversity University (Dinsdale, 1994).
Video teleconferencing: If you were watching a chemist describe the intricacies of a folded protein molecule and its interaction with a drug, chances are much information would be transferred through the hand motions and vocal inflections of the chemist. Body language is one component of physical contact that is difficult to convert to pre-packaged text and still images. Incorporation of these might be accomplished in the future through video teleconferencing technologies over the Internet. The ability to use video teleconferencing could enhance electronic courses not only by transferring these more subtle forms of communication, but also by providing additional visual and audio cues which help the instructor and students to form an informal rapport. Work on the MBONE (Kumar, 1994) and with CUSeeMe (Cogger, 1994) have much potential in this area.
The pencil: One of the most basic visualization tools used in a classroom is the simple pencil. The student uses the pencil to take notes on what is spoken. The instructor will use the pencil to jot down suggestions for future presentations. In our early experiences with providing only electronic copies of the "overhead foils" during a lecture, many students have been frustrated by the inability to take effective notes. Even the annotation capability with the NCSA Mosaic browser is limited by the limitation of annotation only at the full page level, and by the user interface to annotations. We will investigate the CoNote annotation system (Davis & Huttenlocher, 1994).

The creation and use of audio-visual tools is still problematic. The same functionality or at least interoperability must be available across all popular graphical user interfaces (i.e., the Macintosh, Microsoft Windows, and the X Window System). Bandwidth limitations to the students desk also makes many of these tool unusable. It is not acceptable to have to limit participation in the virtual classroom to only those with the best Internet access and interface. Just as html has provided a standard which allows participation by users with diverse implementations of both low-end and high-end browsers, audio-visual tools will require similar standardization.

Pilot Course Description

The initial pilot course offered in our electronic classroom is scheduled for May, 1995. The audience will be interested members of our academic affiliates program which will allow us to get feedback and refine the methods and materials. It is anticipated that the course will be ready to offer to the national community by the summer.

Two specific goals have been set for our efforts in the spring of 1995:

to pilot a course and explore all issues in this remote delivery of materials in the creation of a virtual classroom, and
to provide an introductory level course in concepts in scalable parallel computing, resulting in hands-on experience in parallel programming on the IBM SP2.

We will make use of existing on-line lecture notes, labs, and tutorials by using these in the electronic course. This course will provide educational opportunities to those who cannot attend our traditional workshops held in Ithaca. The materials developed can be re-used by others, especially our Smart Node Program participants. Enhancements made to current materials will serve as a useful reference guide for staff and researchers. Finally the Web community can access these modules, extend them, and incorporate them into computer and computational science courses.

This course will not be an accredited offering. Grades will not be given.

Course Preparation

Preparation for the course has been underway through this winter and spring. Project participants have met to define content and organize materials. Each course module was assigned to a different staff member. Staff review of the materials was done generally by staff from across the Center with a smaller group identified to review in detail. A course instructor has been identified and has the responsibility for integration of the modules, overall coordination, and evaluation. The number of participants will be limited to 20-30. Each student will have a userid on our system for completing lab exercises. An electronic mail alias will be set up to for the students to interact with the instructor. Questions can be submitted while viewing the materials through an html form. As needed, the instructor will enlist other staff to assist with answering questions to ensure a timely response. Questions and answers will be available to all participants through a Frequently Asked Questions (FAQ) document and through immediate additions to the course materials. Because the course materials will be freely available on the Web, the will be accessible by non-participants. Therefore, the materials will be accompanied by a description of the project, the limited nature of participation and inability to directly answer questions from non- participants.

Course Content

The Theory Center's first electronic course, Scalable Parallel Computing on the IBM SP2, is designed to introduce students to basic concepts in scalable parallel computing and all five modules will use Theory Center lecture notes as the starting point. Additional text has been written to fill in the lecture and there are links to appropriate tutorials and labs (which already exist). Links will be added to point to in-depth material on the topic and draw heavily on existing materials such as the Computational Science Education Project (ORNL, 1994) and IBM WWW sites. There will be homework assignments, electronic tutoring, and an open forum discussion for all students and teachers.

Prerequisites: A familiarity with UNIX, C or FORTRAN, graphics capable device and sufficient network bandwidth to download images

Course Syllabus

Module 1 - Getting Started: the Cornell Theory Center Computing Environment: The student will be introduced to the CTC computing environment by familiarizing them with our extensive on-line information system. Specifically, policies, hardware configuration, and training materials will be highlighted. This will be facilitated through an electronic hunt to help them gauge their understanding.
Module 2 - Introduction to Scalable Parallel Computing: This module introduces the student to basic concepts in parallel processing. Performance expectations and limits to scalability are discussed. Students will execute and time existing serial and parallel code to reinforce these concepts.
Module 3 - Message Passing Basics: This module presents concepts in message passing such as latency, bandwidth, computation to communication ratio, barriers and collective communications. It will provide an overview of message passing libraries on the SP2, including MPL, MPI, Linda and Express.
Module 4 - PVM: Since these students also do some of their computing on other platforms, it is critical to instruct in the basics of a portable message passing library. PVM is the most widely used and the basic commands and sample programs will be explored. The student will parallelize an existing program using PVM. Resulting speedups will be shared across the class.
Module 5 - Advance PVM on the SP2: Additional libraries will be discussed and more advanced codes introduced. Students will be required to run two sample codes and create their own parallel program, demonstrating significant speedup through parallelism.

Evaluation and Assessment

This electronic course offering is only a beginning. We will need to evaluate these experiences and build on them for the future. We will ask all participants to complete a course evaluation form. In particular, we are interested in the effectiveness of this virtual classroom for achieving the instructional goals of each module. We also wish to examine how well this virtual classroom can expand to accommodate hundreds of students. We anticipate that many of the materials will be redesigned based on input from participants. We will look at how to expand the course into a full workshop covering a broader set of materials and more advanced topics. We will need to consider how best to work with a larger and more diverse class. An important component of our work will be to share our experiences with others in the Internet community.

Summary

The Theory Center continues to investigate use of on-line information systems for providing effective distance learning opportunities for students and scientists in the areas of computational science and high performance computing. Recent efforts to move lecture materials into html and to incorporate images, animations, and user interaction forms into our documents will provide improvements to our system. Integration of training materials with hardware and software documentation, reports on algorithm design and related research projects will further enhance educational materials on the Web. The transition from providing materials over the network to a virtual classroom will expand our ability to train and educate in our areas of expertise.

References

: Butts, C., Reilly, C., Speh, M., & Wang, J. (1994). WWW and the global network academy. The First Annual International WWW Conference, CERN, Geneva, Switzerland. http://www1.cern.ch/PapersWWW94/speh.ps.; ORNL. (1994). Computational Science Education Project (CSEP) Home Page. U.S. Department of Energy. http://csep1.phy.ornl.gov/csep.html.; Cogger, D. (1994). CUSeeMe. Cornell University. ftp://gated.cornell.edu/pub/video/.; Davis, J & Huttenlocher, D. (1994.) Annotation homepage. Design Research Institute, Cornell University. http://dri.cornell.ed u/pub/davis/annotation.html.; Dinsdale, A.P. (1994). Diversity University Home Page. Wayne State University. http://pass.wayne.edu/DU.html.; Hurley, S., Marshall, A.D., McIntosh-Smith, S.N., & Stephens, N.M. (1994). Courseware for Parallel Computing using Mosaic and the World Wide Web. The Second Annual International WWW Conference, Chicago, Illinois, USA. http://www.ncsa.uiuc.edu/SDG/IT94/Proceedings/Educ/hurley/hurley.html.; Kumar, V. (1994). The MBONE Information Web Home Page. Enterprise Integration Technologies. http://www.eit.com/techinfo/mbone/mbone.html.; NCSA. (1994). NCSA Collage. http://www.ncsa.uiuc.edu/SDG/Software/Brochure/UNIXSoftDesc.html#XCollageNCSA.; Perron, D. (1994). Learning on the WWW: A Case Study. The Second Annual International WWW Conference, Chicago, Illinois, USA. http://www.ncsa.uiuc.edu/SDG/IT94/Proceedings/Educ/perron/perron.html.; Rossman, P. (1992). The Emerging Worldwide Electronic University; Greenwood Press, Westport Conn.