Expert Systems for Experts

Rating:
Summary: Expert Systems by Parsaye
Review: This book contains important details on knowledge-based
systems formulation and implementation. The author describes the
theory of forward chaining which involves matching rules against
a factbase to find successful matches, conflict resolution of
the data and action after concluding a specific alternative
from the various rule structures.

The author describes the expert system life cycle consisting of:
o feasibility analysis of the domain and task studied
o conceptual design of how the expert system carries out tasks
o knowledge acquisition gets the knowledge from human experts
and case sources
o validation
o technical transfer and maintenance
o choice of a manageable task or domain of possibilities

Knowledge is stored in frames with slot names and values.
In backward chaining, the system has a goal to prove and it
works backwards to prove it gaining the evidence with each
iteration. By focusing on conclusions, the system will identify
what caused the pattern of symptoms. The system attempts to
match conclusions in the database until it finds one. In
addition, the system may prove the rule premise. If a matching
rule is found, the system will try to prove each premise in the
rule. Lastly, the AI system may try to backtrack within a premise. If a clause fails, go to the previous clause and
try to find new bindings for variables in the clause. In the event everything fails, find a new rule structure.

This work is a treasure chest of information for anyone seeking
to design a workable AI system utilizing sound programming
and logic principles.