We teach an Introduction to Expert Systems course at SRU. The course counts as an elective in our ABET-accredited Information Systems major, and also counts for Liberal Studies credit in the Math, Science and Technology block. Because we have Liberal Studies (ie non-computer science) students, we needed an expert system shell that was GUI-based and fairly intuitive with a small learning curve. After surveying the available packages, we chose XpertRule Knowledge Builder and have been using it since 2002.
The Expert Systems course is primarily a team-project course where we have teams of students assume the traditional roles of a development team: Project Manager, Programmer, Knowledge Engineer, End User, etc. Typically the team uses someone from outside the class as a Domain Expert. Because the student teams select the project domain, we have had a wide variety of expert systems developed over the years:
Students report the XRKB design interface is straight-forward, intuitive and responsive. The HTML output and reporting features mesh nicely with our web-based development projects. Prior to XRKB we were using CLIPS as the development platform, and there is just no comparison between the two. The GUI interface has the student vote, hands down.
Now these are educational, one-time use systems that get developed in the Expert Systems course. I do not know of any such systems that are in use beyond the scope of the class. However, for a number of years we have been using XRKB in a cross-disciplinary project with Adapted Physical Activity at SRU . This project was funded by a Pennsylvania State System of Higher Education (PASSHE) grant that allowed a professor from Computer Science and a professor from Exercise and Rehabilitative Sciences to develop an expert system to create a safe, but thorough exercise program for community clients.
Every semester the Adapted and Therapeutic Exercise class pairs up SRU undergraduates with clients from our community who have physical disabilities. The typical client is confined to a wheelchair with cerebral palsy. The student reviews the medical records on the client and then conducts a thorough evaluation of the client's physical abilities. Then the student meets weekly with that client in a warm water pool and provides a safe exercise program to preserve and improve the client's physical abilities. Students collect some 130 discrete data values on each client.
Because safety is a very high concern for this program, we wanted to insure the students design an exercise program that is safe and is complete. For example, many clients have physical conditions where one class of exercises would actually be harmful ("contraindicated" in exercise science parlance). A computerized expert system provides a valuable pedagogical tool for this class. Using the professor from Exercise Science as the Domain Expert and hiring a graduate from the Expert Systems course as programmer, we developed an XRKB system that receives all the data points collected by the students and produces an exercise program that the professor in the class might well generate.
Because XRKB seamlessly inputs XML data, we designed a data entry system in VB.NET (with all of its GUI tools). Error checking, completeness testing and the like are all done in the VB system. Then the data is exported both into an HTML format (for printing) and XML (for porting to XRKB). The XML file is then processed through the XRKB system producing an HTML report detailing a safe, thorough exercise program for that client based on the provided data.
So now students in the exercise class develop their own program for their clients, but then receive the results of the XRKB system as immediate feedback. Students then can revise their initial program, bringing in elements of the computer-generated system. Our study has shown that the resultant exercise program used with the client is superior to what the students generate without use of the XRKB system.
We continue to be very pleased with XRKB as a teaching tool in our program and plan to continue its use for the foreseeable future.