Yet another way to control the execution of the rules is to break them into several modules that are called as needed by the main program. Listing Three is an example of this technique. The first function, air_land, decides between driving and flying and returns to the main program. The main program then calls either fly_rul() or drive_rul() to choose a specific vehicle. The module to call next is chosen by the main program on the basis of the information on the global variables.
As always, this sort of modularity improves the clarity of a program. It also increases speed by eliminating sections of rules from consideration.
#include "stdio.h" int av_speed; char like_scenery = 0, is_pilot = 0, fly = 0, drive = 0, fly_com = 0, fly_tcart = 0, fly_bon = 0, m_cycle = 0, car = 0; main() { int done, distance, /* in miles */ time; /* in hours */ char c; char str[10]; printf("This is a program to help you with travel planning.\"); printf("\n\nHow far are you going? (miles)\n"); gets(str); distance = atoi(str); printf("\n\nHow much time do you have for the trip? (hours)\n"); gets(str); time = atoi(str); av_speed = distance/time; printf("n\%d\n", av_speed); printf("Do you prefer scenery over speed? (Y/N)\n"); gets(str); if(str[0] == 'Y') like_scenery = 1; printf("Are you a pilot? (Y/N)\n"); gets(str); if(str[0] == 'Y') is_pilot = 1; air_land(): if(fly) fly_rul(); if(drive) drive_rul(): if(fly_com) printf("\nFly COmmerical."); if(fly_tcart) printf("\nRent a Taylorcraft and fly low."); if(fly_bon) printf("\nRent a Bonanza and fly high."); if(m_cycle) printf("\Take your motocycle and ride the back roads."); } air_land() { if(av_speed > 60) { fly=1; return; } if(av_speed <= 60) drive = 1; } fly_rul() { if(fly && !is_pilot) fly_com = 1; if(fly && is_pilot && like_scenery && av_speed < 199) fly_tcart = 1; if(fly && is_pilot && (100 < av_speed) && (av_speed < 200)) fly_bon = 1; } drive_rul() { if(drive && like_scenery) m_cycle = 1; if(drive && !m_cycle) car = 1; }
In the module air_land(), the first rule causes a return form the module if it fires. This is valid because the average speed cannot satisfy both of the rules in the module. It is desirable because we gain a bit of speed by not considering the second rule of the first one fires.
Another method of accomplishing almost the same thing as separately called modules is to include a state requirement in the IF portion of each rule. Some means must be provided for changing to the next state, typically a line in the THEN portion is some of the rules
Not every rule will need to cause a state change. For example:
if(state== air_land && av_speed < 60) { drive = TRUE state = choose_ground_veh; }
It is also possible to change states at a certain point in the execution of a rule module without reference to the data. In that case, a rule such as this one would be used:
if (state == air_land) state = choose_ground_veh;
This rule would probably be used at the end of a section of rules, organized so that it could not be reached until the state change was appropriate.
Going Further
The ideas here are best used for simple jobs or as a foundation for a study of more complex cases. However, do not overlook the fact that the techniques demonstrated in these examples can be used for a very large but uncomplicated jobs. They could be embedded in a C program to solve a classification problem or replace a complex decision tree.
To learn more about the subject, see the references. You might also want to experiment with the demonstration programs offered by several of the companies that sell expert systems shells. I found the EXSYS demos useful.
Courtesy AI Expert