Home Site

Welcome to PC AI Online

( This Site works best with IE 5.5+ or Netscape 6.0+ )

    Page 55
16.4 Table of Contents Bottom of Page Site Map
asked to pass a student if they answer Q1 as 'a' and Q2 as 'a'. This also fits a passing student's profile as determined from past records.
This technology continuously can update the test database
every time a new student takes the test allowing Knowledge Builder to recreate the decision tree based on this new information. The test itself improves, or learns, with additional profile data.
I developed such an adaptive testing system at Miami-
Dade Community College in 1988. Using records from non-adaptive tests given throughout the semester, it generates an adaptive mid-term exam and an adaptive final exam. For this, I received another award from the League of Innovation.

Recent Advancements
The ability to manage complex, dynamic configurations
is an important advance in some knowledge-based systems. Configuration systems, similar to those frequently employed in manufacturing and assembly, configure products with numerous optional features. For example, every Mack truck is unique - built to order with an assortment of options making it truly unique. To configue a Mack truck for manufacturing, the manufacturer must ensure that they combine the correct engine with the correct transmission; the windshield wipers match the windshield, and so on for numerous other options. The detail level goes down to the threads of the wheel lugs. It takes thousands of constraints for everything to fit perfectly. The choice of one-part type forces the employment of other compatible parts while avoiding those that cause conflicts. The manufacturer must also determine the exact order to assemble the parts. With the Configurator features of a product such as Knowledge Builder, this type of logic is easy to express.
This type of logic is also perfect for configuring training
systems. If you establish the training goals, review the training record of the student, and give the student an adaptive assessment test for placement details, then it is possible for the

knowledge based system to configure an individualized curriculum, different and perfectly suited to each individual student. This curriculum then drives all of the training and testing to achieve the specified training goals.

The Narwhale Student Advisement System
In 1996, I developed an Internet enabled student
advisement system for Miami-Dade Community College - project Narwhale. The system enabled a student to enter a student number to access their student records from anywhere in Florida. On receiving the student record, which included standardized test scores and a list of completed courses with status and grades, the student selected a major. The Configurator (a predecessor of Knowledge Builder) contained the advisement knowledge necessary to advise a student on the courses, the campus, and the times for classes over the next two semesters. With just these two entries, the Configurator generated a proposed schedule more accurately and more consistently than the professional advisors whose advisement knowledge was employed to build the system. Advisors, not faculty, on the Miami-Dade staff created the graphical decision trees that contains the system's knowledge components. The advisors distributed the work between themselves to create the advisement knowledge for over two hundred and fifty majors. The advisors on the Miami-Dade staff were not programmers and they received only four hours of training.

If it is So Great, Why Isn't Everybody Using It?

Knowledge based adaptive training is primarily under-
utilized because it requires slightly more thought than the less effective approaches, and today's instructional technologists are generally unaware of its benefits.
Today's instructional technologists create computer
driven multiple-choice tests with record keeping by using specialized course authoring tools, HTML, and Learning Management Systems. They also create hierarchically arranged tutorials and tests with multi-media effects, etc. using these same tools. This is easy to do since it requires only a surface knowledge of the domain being taught. Most Internet enabled training systems are now built with the emphasis on multi-media and simple statistical scoring rather than on optimal individualized training.
That extra thought and effort to improve the
structuring of the test and tutorial materials, and their interdependencies, is often skipped or not available to the instructional technologist. The training most instructional technologists receive does not teach them how knowledge based systems would enhance their ability to do more efficient training and testing as well as including the multi-media effects that enhance any learning experience.
To create an adaptive system one must add a deeper
layer of knowledge that defines how testing and tutorial materials are related. To utilize this knowledge, the domain experts must acquire the basic skills of working with a knowledge based system development tool or they must work with someone who has those skills. Knowledge based system technology has advanced so far, that anyone with basic computer literacy skills can design, capture, and create the branching logic needed in an adaptive system. Knowledge

To Page 54

16.4 Table of Contents
Top of Page

To Page 56

16.4 2002

PC AI Magazine - PO Box 30130 Phoenix, AZ 85046 - Voice: 602.971.1869 Fax: 602.971.2321
e-mail: info@pcai.com - Comments? webmaster@pcai.com