CRCD
Combined Research-Curriculum Development Program
1997 Grantees Conference Proceedings


Hyperlearning Meter: Toward On-Line Certification

Peter J. Denning
Daniel A. Menasce
Hai Le

George Mason University, 4A5, Fairfax, VA 22030

 

Abstract: Hyperlearning is a term invented by Lewis Perelman to describe new practices of teaching and learning enabled by the Internet and Web. It is best suited to clients who want to develop and certify their competence in a domain -- typically working adult professionals. The traditional classroom model can be characterized as fixed path, fixed time, variable outcome (grade). In contrast, hyperlearning is variable path, variable time, fixed outcome (certificate). The technology for certification is crucial. We have developed a prototype certification system that presents individualized questions to each student, never repeating the exact same question; that generates questions from templates prepared by instructors locally or anywhere in the Internet; and that statistically samples and meters a student's progress against a concept map of the domain. The system also allows students to take preliminary versions of certifications as self-assessment tests; it advises students of their weaknesses and offers links to the modules that will help; and it provides instructors with much valuable data on the progress of a group of students.


Background: The Center for the New Engineer (CNE) at George Mason University (http://www.cne.gmu.edu) has been investigating the philosophy and practice for teaching and learning engineering for the 21st century. We are learning how to serve self-paced, proficiency-based, hands-on learning at the student's desktop without loss of quality. The supporting technologies fit hyperlearning, a non-linear model of learning, that will come to dominate the "virtual university" and commercial educational services. Our first generation of hyperlearning environments are embodied in a library of eleven CNE modules. The most difficult technical problem is assessment of student learning.

You will find it helpful to sample our library of modules on the CNE web pages. Pay special attention to the subway maps used for general navigation and to the self-assessment tests embedded in the math and statistics refresher modules.

In 1994, as new CRCD investigators, we undertook to develop online modules that would teach engineering students about complex systems in other domains that had research tools and methods of use to engineers. By the end of 1995 we had built a small library of 11 modules that were used successfully in computer science classes. These modules enabled instructors to spend less time on certain topics during class; and they enabled the students to complete more sophisticated homework problems.

By the middle of 1995 it was clear that distance learning was a central focus for many educators contemplating future curricula and that our library of modules represented an attractive architecture for those interested in asynchronous distance delivery of courses. The structure of the modules, we discovered, closely matched the hyperlearning structure envisaged by Lewis Perelman in his book, School's Out (Avon, 1992). At the same time we engaged in a project with the Defense Acquisition University to use our hyperlearning architecture to create math and statistics refresher modules for their cost accounting students. On-line self assessment testing was an essential ingredient of these new modules. This is when we undertook the design of the certification and assessment system outlined here.

Models of Learning. A traditional course is a sequence of topics covered in a series of lectures, held in classrooms at weekly intervals, with homework practice in between. This is a linear model of learning. Everyone proceeds at the same pace regardless of their interests, prior experience, talents or other demands on their time. At the end, grades indicate the level of achievement a student was able to make in the fixed time period allocated for the course. We summarize the traditional classroom model as fixed path, fixed time, variable outcome.

Imagine a new model. Instead of a classroom, see in your mind a large "learning room" with an entrance, an exit, and a number of learning stations (booths). You meet the teacher on entry. The exit is guarded by a certifier, whose job is to assess your competence against well-defined standards. You visit the stations to learn particular topics or practices. Colored lines on the floor suggest paths among the stations. You can visit as many stations as you need, and in any order consistent with your current knowledge, to prepare yourself for final certification. You can take trial certifications and then backtrack to the stations needed. You can take self-assessment tests at any time you like. You call on the teacher for help at any moment you are stuck. The teacher will offer guidance if you are heading in a wrong direction. In contrast to the linear model, everyone who exits gets the same "grade" (a certificate of competence); the variables are the length of time and the path followed. We call this the hyperlearning model. The prefix "hyper" means non-linear, the ability to jump to other dimensions, as in a mathematician's hyperspace or an author's hypertext. We summarize the hyperlearning model as variable path, variable time, fixed outcome.

The CNE Modules. The hyperlearning model is easily implemented in the World Wide Web: a learning module contains information objects representing the stations, self-assessment tests, texts, pictures, demos, workbenches, links, and certifiers. For navigation, we use a subway map whose colored lines connect the stations and lead to the certifier. Our library includes eight modules in computer science subjects, one module on the general engineering subject of senior design, a refresher on high-school mathematics, and a refresher on introductory college statistics. The computer science modules are used in classes, where they allow instructors to shorten the classroom time on those topics and the students to do more sophisticated projects. The math and stat refreshers are being used by the Defense Acquisition University. A Module Authoring System permits teachers to construct their own learning modules. The CNE modules site receives 4,000 to 10,000 "hits" per week and has been awarded a three-star Magellan rating.

Module Authoring Systems. One of the objectives is to make the modules technology available to course instructors. The central concept for navigating a module is the subway map; the map organizes all the files in which module segments are stored. We are developing a subway map editor that allows an instructor to (a) draw a map representing the conceptual structure of the domain, (b) create sensitized icons representing subway stations that are linked to the corresponding module segments, (c) arrange the lines, stations, and colors for a pleasing and usable appearance, and (d) perform a consistency check on the links connecting all the components of a module.

Certifiers. Certifiers confirm that a student has met the learning objectives of a course and then issue an authoritative declaration of the student's competence. The first generation of certifiers deal with problems whose answers are algorithmically computable, such as occur in math, science, and engineering courses. The technical challenges include: designing test generators that accurately assess students according to given criteria, validating the tests, providing a data system giving instructors feedback about student performances on self-assessment tests, scrambling questions and multiple-choice answers to prevent fraud and allow students to meaningfully retake tests. The certifier generator should generate tests that completely cover the target domain, and it should provide feedback to the students about correcting their weaknesses. A Certifier Authoring System allows instructors to create certifier templates containing text, formulas, graphs, images, video clips, applets and sound. The certifier system consists of these components:

(a) an editor that allows instructors to prepare templates for test questions using both correct and incorrect algorithms,

(b) a program that processes a template into a particular exam by substituting numerical values for parameters and computing results according to algorithms supplied by the instructor,

(c) a database containing all templates and student records, Even when tests are used for self-assessment, the database will record a student's progress and can give the instructor information about strong and weak areas of knowledge both for individuals and for an entire class, and

(d) a concept map editor that allows instructors to organize the concepts of a course into a network of interrelated concepts. The concept map is also used as mechanism to report on the progress of students.

Overall Conclusion. The original purposes of the project were met by the termination date at the end of December 1996. We demonstrated that we can teach students concepts and principles from complex systems using on-line hyperlearning environments. We found that instructors can use the modules to augment class discussions, decreasing the amount of class time required on those topics and increasing the students' capacity to solve complex homework problems. We found that the self-assessment tests are a powerful aid to student learning and incorporated them as a standard part of every new module. We constructed a prototype certification system that will be a standard part of every hyperlearning environment.

Acknowledgments: This material is based upon work supported by the Combined Research-Curriculum Development Program as National Science Foundation Award EEC-9315476, and by DARPA Contract DABT63-93-C-0026. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation or of DARPA.