Procedure for production
Initially we have a data base that has updated data and we continue to work with operators till we reach a goal condition.
We select some rule’ R’ from the set of rules applied to the data.
We apply that rule to generate new data.
We continually perform this until we reach the required condition.
If the required condition is achieved then we say that the problem is achieved.
Control Strategy
Selecting the rules and keeping track of those sequence of rules and the database they produce constitute is termed as control strategy for production system.
Operations of AI production systems can be characterized as a search process in which rules are tried until some sequence of them is found that produces a database satisfying the termination condition.
Computational Cost
Very important attribute for selecting rules is the amount of information that you aware of about the problem.
In the worst case if I don’t know any thing about the problem the rule can be selected arbitrarily, as I don’t know anything about the problem.
At the informed extreme , when I know the complete knowledge about the problem the control strategy is guided by the problem knowledge and this guidance is enough to select a correct rule every time.
Overall computational cost of a AI production system can be categorized into two major categories.
Rule Application Cost:
For uninformed Control System:
In this system we have try a large number of rules to find a solution ,Therefore we have a high rule application cost.
For informed Control System:
In informed control system the production system is directly guided to the solution, Therefore very less Rule Application Cost
Control Strategy Cost:
For uninformed Control System:
Here in this case arbitrary selection does not depend on costly computation. Therefore we have a low computational Cost.
For informed Control System:
To inform about the problem domain, cost in terms of storage and computation. Therefore we have a high control strategy computation.