Distance learning using embedded artificial intelligence

by MAJ Jeff Girard

Distance learning, on-the-fly-training, just-in-time learning, computer-based training and computer-based instruction. These are the buzzwords of the education and automation world for this decade. Regardless of what label one uses, the concept is the same: provide instruction and/or instructional material to a student not necessarily collocated with the source material and/or the instructor.

Distance learning is really an umbrella term that encompasses everything from mailing out instructional material for courses (correspondence-course program) to the innovative use of the Internet to bring together groups of students into one virtual classroom using simulations. DL is not a "one size fits all" solution. The system I’m about to describe is one part of that continuum designed to meet a very specific audience’s needs.

The system I’ve built using worldwide web technology provides individuals a way to access the enormous amount of data housed within Training and Doctrine Command’s school system. These data resources are the digitized training-access centers that are an integral part of TRADOC’s Classroom XXI design. This system provides for the intelligent retrieval of information, data and training resources. This intelligent retrieval system means that only information that’s necessary to meet the user’s needs is provided. The system tailors each individual resource package to meet that customer’s needs. The files are then compressed together into one file that can be downloaded to the customer’s computer system. The customer is then able to disconnect and process through the training resources at his or her leisure.

This system was designed to provide the customer with an intelligent resource retrieval system. It wasn’t intended to provide complete courses and/or any type of certification. It was intended to meet the needs of customers who desired more or refresher training and professional development, or who needed to be brought up to date on a particular subject area.


Throughout this article, I’ll use my own experiences as an illustration. I relinquished command of D Company, 13th Signal Battalion, 1st Cavalry Division, in December 1994. I then entered a four-year gap where I wasn’t directly involved with the tactical Signal community, which ended when I joined 10th Mountain Division (Light Infantry) at Fort Drum, N.Y., in June 1998. During my "sabbatical," there were significant changes in the doctrine and equipment involved with providing tactical Signal support to a division fielded with mobile-subscriber equipment. The Force XXI experimentation at Fort Hood, Texas, incorporated routers and high-speed multiplexer cards in Signal assemblages. Also, my entire Signal-support experience had been in heavy mechanized divisions. Now I’m assigned to a light division with the force-entry switch and dismounted-entry switch. I’ve forgotten all I ever learned about that equipment years ago. What am I going to do? I’ll be expected to know about this equipment, the current doctrine and how to support the light division if I’m ever to be successful at my job.

The Signal Center at Fort Gordon, Ga., would be the most logical place for me to look for help. But what do I do? I could use a search engine and try to conduct a search on these topics, but that would only be marginally helpful at best. I could take a lot of time and follow all the links on the Fort Gordon pages, but that may result in a lot of hours spent on the keyboard with no positive results. There are no resident courses I could take to teach me on the specific material I need, nor are there any correspondence-course materials for me to use. I need a system that will help me retrieve the materials I need from resources that may be located anywhere in the world. I need an automated, intelligent counselor who, upon listening to my problems and my requests, knows exactly the materials I need, will electronically retrieve them for me and will deliver the same materials to me at my request.


The first and crucial step in designing any software solution is an accurate identification of the problem. As outlined above, I’ve defined the problem as a need for an intelligent software package that conducts an analysis of the query and spawns an "intelligent mole" that executes database mining to identify and retrieve only data that solves the query. Also, the system must be accessible to users worldwide, which indicates the system must be accessible via the Internet.

The solution to this problem is to use education, database-retrieval, WWW and artificial-intelligence techniques to create an "intelligent librarian." This solution frees the user from having to conduct searches of the wealth of data available on the net. This system uses adaptive-questioning techniques to identify the delta between the user’s current basis of knowledge and what the user desires to know. The system then conducts an intelligent search of databases and other repositories of electronic media, which will enable the user to gain the knowledge identified as the delta. The system then retrieves copies of this data, compresses it together into one file (along with any required viewers and plug-ins) and then provides that file to the user for consideration. Finally, this system is based upon WWW technology that makes it accessible to anyone with a browser-capable computer system.

The figure below indicates the circular nature of the adaptive-questioning techniques. When the system completes its analysis and has determined it has identified the knowledge delta, it initiates the resource-collective phase. These multiple resources are then compressed into one file, along with any required readers and/or plug-ins for the particular media type, and delivered to the user via an http download.

Adaptive questioning process for distance learning Adaptive-questioning process for determining user's knowledge base.

Scenario: analysis phase

As outlined previously, I realize I need assistance in catching up with doctrine and technology in the Signal Regiment, with particular emphasis on Signal support to light divisions. I turn on my browser-capable computer system and point my browser to www.somewhere.mil (or .gov, .org, .edu or .com as the utilization dictates). I’m presented with a series of webpages that provide some instructions for first-time users and a menu of subject areas from which I select "MSE support in light divisions" (figures below). This selection activates an expert system that runs in the webpages’ background. I’m presented with a series of questions specifically designed to establish my current basis of knowledge. In this particular example, I’m asked about node-center switches, extension nodes, data rates, engineering criteria, etc. I’m also asked questions that are specific to the capabilities and uses for FES and DES.

Welcome page for distance-learning system Welcome page for distance-learning system.
Instructions for distance-learning system Instructions for distance-learning system.
Main menu for distance-learning system Main menu for distance-learning system.

Throughout this analysis phase, adaptive-questioning techniques are used. Correct responses to questions indicate a higher level of knowledge and therefore will generate a more difficult question, while incorrect responses will generate easier or more basic questions. Concurrently with the cyclic questioning, the expert system is firing rules that identify those specific subject areas in which I need more training. The expert system is constantly modifying the rules list depending on the cyclic questioning’s results and the adaptive-questioning techniques used. The cyclic-analysis phase’s endstate is that the system has accurately determined I’m knowledgeable about MSE communications equipment used in a heavy division, but I’m not familiar with the equipment used in a light division. The system also concluded I have an adequate knowledge level of the electrical theories used in engineering MSE systems.

This is my knowledge base for the subject in question, and so the system concludes the analysis cycle and begins the collection-and-delivery process.

I wasn’t able to include screenshots with this article from the analysis phase, unfortunately, as I wasn’t able to obtain a release from the vendor of the expert tool I used to develop the expert system that runs in the background.

Scenario: resource-selection process

The system has determined I have an adequate knowledge of heavy-division equipment and I know the electrical theories used in engineering communications systems. The expert system has also determined my desired endstate is to be knowledgeable about the same subjects with respect to the equipment and doctrine used in a light division. As I said, the expert system continually fires rules, which update the list of subject areas in which I need assistance. The delivery process begins by retrieving this list from the expert system.

The system uses this list to retrieve specific training resources. In the scenario, the system would retrieve these resources from all the digitized resources available in the Signal Center’s DTAC. The system would draw text files, paragraphs/pages/chapters from field manuals and training manuals, photographs, videoclips and/or audioclips from DTAC to build a training-resource package that would be specifically tailored to meet my particular needs.

Depending on my needs, this collection of resources may be quite large. Therefore the system provides the user with the opportunity to further refine the tailored training package. The system presents a webpage to the user listing all the recommended resource files (top figure below). Each of these file listings is actually a hypertext link. By following the link, the user is presented with detailed information about that individual file. This detailed information includes the subject covered by the file, the file size, file type and any required drivers needed to view the file: for example, Motion Picture Experts Group viewers or Real Audio (bottom figure below).

File check for distance-learning system File check and check-out for distance-learning system.
Response screen from distance-learning system The user is presented information on files gathered by the distance-learning system.

Scenario: resource-download phase

This system was designed to support users worldwide using a variety of means to access the Internet. I took this concept into account when I developed the download process. When the user has selected all resource files he or she wants by checking boxes on the webpage (top figure above), the system retrieves all the resource files, along with any required viewers and/or plug-ins, and then compresses them into one file using a compression utility. I anticipated this file might be quite large to download, so I present the user with several options. I tell the user the file’s size and an approximate amount of time it’ll take to download using a variety of means. Finally, I offer the user the opportunity to download the file, come back later to download the file or exit out of the program without saving any information (figure below). One choice I provided to the user is to come back at a later time (during off-peak hours, during the night) to download the resource file. I used a system of webcookies to accomplish this task.

Download and exit for distance-learning system Download and exit options for the distance-learning system.

Scenario: program conclusion

Regardless of which option the user selects, the program terminates cleanly. The figure below shows the terminating webpage the user will see. The page offers the ability to send comments and questions to the author, as well as the ability to restart the program to initiate a new session. The user doesn’t see that the program also conducts some housecleaning functions as it deletes temporary files established as part of creating the user’s tailored resource package.

Terminating page for distance-learning system Terminating page for the distance-learning system.


In the age of a downsized Army, soldiers will be continually expected to perform more skills with less formal training. One of the negative results of this combination is that soldiers are being required to perform duties for which they have no prior formal training. To complicate matters, the formal educational institutions aren’t resourced to provide enough training, so the armed forces are left with a training deficiency and no immediate solution in sight. Soldiers will be expected to participate in professional-development courses, computer-based training and other individualized training. The "new millennium" soldier will be expected to assume the personal responsibility of training himself or herself when the formal education systems within TRADOC are simply insufficient.

I’ve demonstrated to you a system I’ve built and that actually exists. This system is web-based and is available to anyone with a web browser via the Internet. The system provides tailored packages by using an expert system which assesses the student and creates a package specifically tailored to meet the student’s individual needs. The system is boundless in that this manner of training is only limited by the ability of instructors to put material in a web-based format. Most importantly, this system is available now. This DL method is available today by making use of existing technology and is of an open-architecture design to grow in the future.

The uses for this type of system are limitless. Some potential applications are:

Soldiers who are performing a function for which they need more training;
A soldier’s chain of command ensuring that soldier meets a certain proficiency level before his or her arrival at a resident course of instruction; and
Signal Center leaders dispersing materials worldwide for low-density educational needs.

The National Guard and Reserve forces would be the biggest benefactors from such a system; they would have almost limitless training available to them at their home stations. This would greatly reduce the costs associated with sending soldiers to resident courses of instruction.

Technology and the capabilities being provided by automation are growing exponentially. One of the advantages this nation enjoys is that we consistently take advantage of those capabilities to give us that "edge," whether it’s in an economic, political or war setting. We must continue with this established methodology and make full use of the advantages technology is providing to us. I’ve demonstrated to you how we can develop a singular system that can provide for our training deficit being created by the downsizing of our Army.

MAJ Girard is 10th Mountain Division’s division automation-management officer at Fort Drum, N.Y. His previous assignments include a year as 10th Signal Battalion’s executive officer; out of these 12 months, the battalion operated for eight months as split task forces to support Bosnia and the advanced warfighting experimentation preparation at Fort Drum. He has a master’s degree in artificial intelligence from Duke University and has been developing AI applications since 1984.

Acronym QuickScan
AI – artificial intelligence
DES – dismounted-entry switch
DL – distance learning
DTAC – digitized training-access center
FES – force-entry switch
MSE – mobile-subscriber equipment
TRADOC – Training and Doctrine Command
WWW – worldwide web

dividing rule

Back issues on-line | "Most requested" articles | Article search | Subscriptions | Writer's guide

Army Communicator is part of Regimental Division, a division of Office Chief of Signal.

This is an offical U.S. Army Site |


This is an offical U.S. Army Site |