Classroom 2000: Designing the Student's Assistant

Gregory Abowd
Graphics, Visualization and Usability Center, College of Computing, Georgia Institute of Technology, Atlanta, GA 30332-0280,

Proposal submitted to the Hewlett-Packard Research Grants Program

February 26, 1996


With the availability of ubiquitous information technologies, most notably the World-Wide Web, most universities are able to provide students with access to vast repositories of educational materials. The dizzying number of Web pages devoted to classroom materials, from all disciplines, is a testimony to the value of these services, at least from the perspective of the instructor (for example, refer to the World Lecture Hall site). But ubiquitous computing in an educational setting is about more than just bringing the classroom, or more accurately a static view of the classroom experience, to the student. In work done at Georgia Tech, we are concerned with developing futuristic ubiquitous computing technology to support the entire classroom experience and experimenting with it today in order to understand how it will change our approach to education for the better.

For example, one of the interesting issues we would like to address is what will be the impact on the classroom when every student brings a personal digital assistant (PDA) or notebook computer to class? These computers will have wired and wireless connections to the campus network, allowing them to be used for communication as well as computation. Will these enviable resources be used to entertain students during dull lectures, to enhance current teaching approaches, or drive an evolution of new approaches to learning? Will we have to ask the students to put the machines away in order to preserve group interaction? Can we use these resources to enhance interaction? Will we still lecture to students, or will new forms of pedagogy evolve? Will we have formal class meetings at all, or can we use enhanced email and news groups to mediate some of the classroom interaction?

Our aim is not to replace the traditional lecture-based style of pedagogy, at least not initially. But we do feel that a lot of the information in a lecture is lost and inefficiently recorded. The long-range goal of Classroom 2000 is to make it easier for both teacher and student to benefit from the rich exchange of information that is possible in a classroom setting. We will move toward that goal by developing a suite of tools and a development methodology that allows us to build a software infrastructure for supporting the delivery, capture and review of educational materials.

It is important to contrast our approach with the more traditional approaches to introduce technology in the classroom. Much of this work has been summarized by Shneiderman in which teachers are provided with state-of-the-art presentation facilities and students sit in front of workstations. Whereas this approach to classroom technology has been shown beneficial to courses which involve programming, it is not suitable for classes in which programming is not the essential subject matter. Many courses are also videotaped for later viewing, but the majority of those videotapes remain on shelves as students are not willing to replay the entirety of a lecture searching for the relevant parts.

Overview of proposal

The remainder of this proposal is structured in the following way. We will first summarize our research objectives for this work. We will then summarize the progress we have made so far in meeting the objectives of Classroom 2000. We will outline a 3-year plan for meeting the remaining objectives and describe tangible milestones that will mark our progress on a yearly basis. We will conclude with a suggested budget for personnel and equipment, a discussion of the suitability of research facilities at Georgia Tech for this particular project and a biography of the principal investigator.

Research Objectives

We are interested in prototyping future computing environments that will enhance the classroom experience and empower both teacher and student. In this proposal, we describe a research effort, the Classroom 2000 project, which will investigate the introduction of ubiquitous computing technology in the classroom. We have several research questions that we use to formulate objectives for our work.

In answering the first question, we look to many of the concept videos produced by large industrial firms (Hewlett-Packard, Apple, Microsoft and IBM) and the visionary work of researchers such as those at Xerox PARC. The predictions of the world in 10-15 years emphasize that the computer interface of the future will be everywhere. We can uncover many common themes for our future computing environment: ubiquity and mobility of the interface; high bandwidth connectivity; physical context awareness; and automatic machine learning. Much of this capability is available to us today, but it is not being effectively packaged to solve problems. The specific technology we want to employ in Classroom 2000 involves pen-based technology on large electronic whiteboards and on personal handheld or laptop notebooks. The traditional classroom lecture usually involves an instructor using visual aids such as overhead slides and notes written on a publicly displayed whiteboard. The students struggle to pay attention to what the teacher is saying and also write down the salient information on their own notes for review later. This is an ideal application for promoting pen-based technology. Replace the whiteboard and student notebook with electronic counterparts and digitally record what is going on during the class and we can now capture a lot of the information that the student needs for review purposes. And all of this technology is available today, some of it even at a fairly reasonable cost.

This immediately leads us to the second question of how to introduce the new technology into the classroom in order to evaluate its effectiveness. Here we draw an important distinction between our work and other work in technology to support education. Much of the research we are proposing here does not represent a radical shift in the way teaching will occur. We are still assuming traditional lecture-based forms of pedagogy and we are looking for ways to incrementally improve on that through technology. It is also important to look at new ways to teach, but rather than suggest revolutionary teaching methods supported by technology, we are interested in seeing how traditional teaching methods evolve when augmented by that technology.

The final research question focusses on the practical issues involved with building heterogeneous systems. We have admitted that a lot of the technological solutions for the work in Classroom 2000 are available, but they were not developed to work together to realize the intelligent assistant that we envision. We need to come up with generic architectural solutions for preparing, capturing and filtering information for use in the electronically enabled classroom environment. Our research approach is to build different versions of functional prototypes and determine through experience what common abstractions they support. Through the experience of building the heterogeneous and complex system we will gain knowledge of the useful structuring and coordination devices that define generic architectural solutions. And generic architectural solutions are a precursor to tools which help to generate families of software systems with common functionality.

To summarize, our research objectives are:

Current accomplishments

In September, 1995, we received a seed grant to initiate work on Classroom 2000. In the ensuing six months we have completed our first prototype classroom and are currently evaluating the effectiveness of this technology as it is used live in a graduate class in the College of Computing at Georgia Tech. In this section, we will highlight the features of the system we have built. Further details are available on this project, including access to the information accumulated in our live evaluation class. Our work in Classroom 2000 has been greatly influenced by the work in ubiquitous computing[6] and electronic notetaking[5] done at Xerox PARC. We want to capture information provided by the teacher during a lecture, so electronic whiteboard capabilities provided by the Xerox LiveWorks LiveBoard [1] naturally suggest themselves. The teacher in Classroom 2000 uses the Liveboard and our software to present and annotate prepared lecture presentations.

We also wanted to provide the students with an electronic notebook with the capability to take notes during the class that could be the basis for review after class. The Marquee note-taking prototype developed at Xerox PARC [5] came the closest to what we wanted to have in the hands of the students. Marquee provided a simple mechanism for producing notes with a pen-based interface that also created automatic indexing into a video stream. We currently have the capability to handle digital audio recordings of classroom lectures and will eventually be able to support digital video. Using the students's electronic note-taking as a simple and automatic way to index into those rich information streams provides a value-added capability beyond paper-based notes.

Framework for Classroom 2000

In designing the prototype for Classroom 2000, we found it useful to divide the activities into three distinct phases: pre-production: in-class use; and post-production.


In the lecture-based model of the classroom, we assume that the teacher does some preparation for each lecture. This pre-production phase can range from the preparation of a complete slide-based presentation that will be shown in class to a less formal preparation of notes that the teacher alone will use during the lecture. Any prepared materials that the teacher wishes to make available to students during the lecture must be transformed into a format that can reside on the student's electronic notepad and be presented on the Liveboard during class.

In-class use

We developed a single software system, ClassPad, shown in Figure 1, that runs on both the Liveboard and the student notepads. During class, as the teacher is presenting the lecture, she can mark the slides with additional annotations. Similarly, the students can annotate their own private copy of the slides. Blank slides can also be created on the fly in order to insert additional notes that might not fit on the prepared slides. The entire lecture is currently audio recorded and stored on disk; we will soon be able to support digital video recording as well. Every pen annotation is recorded with a timestamp to facilitate audio indexing. It is important to stress that in designing the ClassPad application, we tried to introduce as little extra work for the note-taking student.

Figure 1: Screenshot of ClassPad prototype


Once the lecture is complete, we enter the post-production phase of Classroom 2000. We have tried to treat the whole data collection and reviewing task as a Web authoring and browsing task. A class home page holds a syllabus containing all information accumulated for each lecture, including the prepared lecture notes, the teacher's and student's annotations on those notes and the audio track. The ClassPad application accumulates annotations and timestamps and produces an HTML document that can immediately be hung off the class Web page for browsing. The browsable form of the notes contains links attached to each slide that accesses a customized audio service tool to deliver the portion of the lecture pertaining to that slide. A sample of the browsable notes page is shown in Figure 2 below.

Figure 2: Screenshot of Web-browsable class notes.

We are also investigating alternative interfaces with content-based search facilities. Figure 3 below shows a searchable interface that has several advantages over the original. First, we provide thumbnail images of the notes from one lecture across the top to enable easy access in a non-linear fashion. The original prototype only allows the user to scroll down from the first slide to the last slide. Second, the currently viewed slide is shown in a separate frame with the audio links in an adjacent frame. Finally, the student can perform a keyword search across all lectures to produce a presentation that covers a single topic.

Figure 3: Alternative searchable interface for class notes.


We have evaluated the prototype environment since mid-November 1995. The first two months of use were in the context of an Object-Oriented programming course, in which we used the Liveboard extensively and received some preliminary feedback on the initial handheld ClassPad prototypes. We are currently teaching a class on Human-Computer Interaction which is the first live application of the Classroom 2000 technology. We have taken a formal pre-course questionnaire to understand the students' impressions of the technology coming into the course. We do not have enough handheld units to provide each student (there are 27 students in the class and we have a maximum of 12 handheld units depending on hardware availability), so we are asking students who do not use the electronic note-taker to keep a journal containing their handwritten notes for comparison with the electronic notes. Students also fill out a daily log to indicate their reactions to the use of the technology in that class and in the period of time between the previous class and the current class. We will conduct a formal post-course questionnaire to assess their reactions after the fact. We are also logging use of the class web pages, with particular attention being paid to the class notes and the audio links.

Research Plan and Schedule

We have made significant progress in the Classroom 2000 project over the past 6 months, and we can now determine the path we would like for this research to follow for the coming years. We want to view this as a multi-year proposal, but since Hewlett-Packard will only guarantee funding on an annual basis, we will pay particular attention to the first year's plan.

Year 1 plan

Better audio support. The current implementation provides a client-side audio player for stream-based playback of Sun AU audio files. This is a much better approach than the current packet-based audio players provided as helper applications or plug-ins for current browsers because it allows the student the freedom to move anywhere in the audio file. However, the current audio player runs only on SunOS or Solaris workstations, and this has proven to be a drawback for students who do not have regular access to such machines. We are currently implementing a Java-based audio tool that will provide us true cross-platform audio playback support.

Content-based search. The current prototype only allows searching based on the contents of the prepared slides. It would be more useful to allow content-based search on the information that occurs during class as well, such as the student/teacher annotations or the recorded audio. We will be investigating techniques to search based on analysis of the digital audio as well as pattern recognition of the pen annotations.

More live classroom experience. We will be evaluating the Classroom 2000 prototypes in more classes. We want to gain experience with a variety of instructors, students (graduate/undergraduate, computer science/liberal arts), and teaching styles (lecture/project/discussion courses). Though we are not specifically aiming Classroom 2000 at a particular discipline, we do not assume that our approach will be suitable for all possible combinations of the above variables. It is just as much a contribution to determine which situations are unsuitable for Classroom 2000.

Variability of ClassPad platforms. We mentioned that one of our objectives was to be able to provide some architectural insight into how to develop heterogeneous hardware/software applications such as Classroom 2000. The current ClassPad prototype was built for a PC running Windows for Pen Computing. We are currently building a prototype for Newtons. We would also like to develop prototypes using other pen-based platforms, such as that used on the HP OmniGo.

Increased connectivity. To better facilitate interactions inside the classroom, we want to provide wired Ethernet capabilities to each student notebook. This will enable different kinds of information exchange within the classroom, such as anonymous feedback to the instructor, on-line questioning, shared whiteboard and in-class Web browsing (to see previous lecture's notes, for example).

Improved post-production capabilities. The current post-production activities assemble a single HTML page for review based on a log of time-stamped annotations. There is no facility for the student to further revise the notes except by manually editing the HTML files that are generated. Further support for these post-production/review activities will become even more important as we try to evaluate the effectiveness of the technology across many different courses. We have some ideas for post-production capabilities and have prototyped some of them, but this issue has largely been ignored up to this point.

Year 1 milestones

The following deliverables will serve as milestone achievements for the first year's work in this proposal:

Year 2 plan

Video support. We will build or obtain suitable stream-based video player analagous to the audio player and investigate content-based search techniques suitable for video.

Wireless connectivity. To support student activities outside of the classroom, we will have to provide a wireless solution.

Very large interactive whiteboards. We want to provide electronic whiteboards that are the size of normal clasroom whiteboards and we believe we can do this a number of ways, depending on budget constraints. The important aspect of the electronic whiteboard is that prepared material is displayed and annotations are displayed and logged for later playback.

Dynamic information support. In the same way that we currently provide the student the ability to annotate and replay static information, we want to support the annotation and review of this more dynamic information, such as video segments, computerized simulations or animations, or physical demonstrations.

Year 2 milestones

The following deliverables will serve as milestone achievements for the second year's work in this proposal:

Year 3 plan

Building the intelligent student assistant. We will use machine learning techniques to track activities of a single student in order to predict how to better assist the note-taking and review process.

Evaluation across multiple classes. We will begin to develop strategies for deploying Classroom 2000 for individual students and track them across multiple classes over multiple semesters/quarters.

Year 3 milestones

The following deliverables will serve as milestone achievements for the third year's work in this proposal:



We will employ one graduate research assistant with funds from this proposal. The cost of a single College of Computing graduate research assistant for a full year is $16,000 for salary and 1,670 for computing charges.


One of the most important questions we are trying answer is the effect that Classroom 2000 technology will have when in the hands of every student in a class. To do this, we require 50 portable pen-based computers to evaluate the Classroom 2000 concept in a realistic setting. This would allow us to provide each student with an electronic notebook during the class. Since we will want to explore use of this technology across several classes, this number of units would not allow us to give the computers to the students for use outside of class. However, should this proposal be accepted, we would like to enter into a more meaningful discussion with Hewlett-Packard concerning the availability of suitable equipment for Classroom 2000. We are aware of the OmniGo, but do not know of any other pen-based equipment that HP retails. In conversations with Fredrick Kitson, of HP Labs in Palo Alto, we are aware of some other prototype portables that HP is working on. We have seen a presentation by Joel Birnbaum that discussed an electronic notebook being developed by HP that seems to be ideally suited for our purposes. We would like serious consideration of the possibility for us to use this research technology.

For development purposes, we would like 2 high-end HP workstations; the particulars of these development machines will depend on the portable strategy we employ. We would also require a server machine to host all information collected during Classroom 2000. We would like to go with an HP Netserver machine with enough disk space to serve what will likely be in the neighborhood of 2 Gigabytes of information per class served. In future years, we would like to investigate what HP has available in the area of large screen display technology and wireless communications services. We are unaware at this point whether these are available HP products.

About Georgia Tech

We are in an enviable position at Georgia Tech to pursue projects such as Classroom 2000. Professors Abowd and Atkeson in the College of Computing have initiated a research effort entitled the Future Computing Environments (FCE) Group with the aim of building proof of concept prototypes for ubiquitous and mobile computing. The Classroom 2000 project was initiated by the FCE group as part of an internal interdisciplinary university seed grant. The Graphics, Visualization and Usability (GVU) Center is a premier center for multi-disciplinary research which is essential for developing and evaluating the effectiveness of technology. The EduTech Institute has recently been established to support research into the specific questions concerning the impact of technology in education. Researchers from EduTech are currently helping us to assess the effectiveness of pen-based note-taking and Web delivery of audio-annotated review notes in Classroom 2000. The state of Georgia is sponsoring a Georgia Tech initiative on broadband communications applications, entitled FutureNet, which is providing a lot of the infrastructure necessary for wired and wireless broadband communications to classrooms and student living quarters. FutureNet is now looking for innovative services to provide across this infrastructure and has become involved in the development and delivery of audi services for Classroom 2000 and we are currently working on providing similar services for digitial video delivery.

Biography of PI

Gregory D. Abowd

Gregory D. Abowd is an assistant professor in the College of Computing at Georgia Institute of Technology, where he is a member of the Sofware Systems Design Group, the Graphics, Visualization and Usability Center and is director of the Multimedia Lab. His research interests include software engineering for interactive systems, with particular focus on future computing environments. He has recently started a research group in the College of Computing focussed on the development of prototype future computing environments which emphasize ubiquitous and mobile computing technology for everyday uses. Dr. Abowd received a BS in Mathematics from the University of Notre Dame in 1986 and the degrees of M.Sc. (1987) and D.Phil. (1991) in Computation from the University of Oxford, where he attended as a Rhodes Scholar. He is a member of the IEEE Computer Society and the ACM. Before coming to Georgia Tech in 1994, Dr. Abowd held post-doctoral positions with the Human-Computer Interaction group at the University of York in England and with the Software Engineering Institute and Computer Science Department at Carnegie Mellon University. Dr. Abowd has published over 40 scientific articles and is co-author of one of the leading textbooks on Human-Computer Interaction.


  1. S. Elrod et al. Liveboard: A large interactive display supporting group meetings, presentations and remote collaboration. In CHI'92 proceedings, pp. 599--607, May 1992.
  2. Alex Poon, Karon Weber, and Todd Cass. Scribbler: A Tool for Searching Digital Ink. In CHI'95 proceedings, Short Papers. March 1995.
  3. B. Shneiderman et al. Windows of opportunity in electronic classrooms. Communications of the ACM, 38(11):19--24, November 1995.
  4. Nitin "Nick" Sawhney, Gregory D. Abowd and Chris Atkeson. Can Electronic Notebooks Enhance the Classroom? GVU Technical Report GVU-96-xx, February 1996.
  5. K. Weber and A. Poon. Marquee: A tool for real-time video logging. In CHI'94 proceedings, pp. 58--64, April 1994.
  6. M. Weiser. The Computer for the 21st Century. Scientific American, September 1991.