Second Generation Expert
Systems
(motto: “knowledge is power”)
Problems, computer, resolution, algorithm, AI
“Knowledge!”
“What is knowledge?”
What is the difference between information and knowledge?
Required characteristics of knowledge
“Is there any knowledge in computer programs? Where?”
I. generation
(Classical)
Expert Systems
“What is the difference between Expert System and Knowledge Based System?”
Knowledge base + inference engine
“How to construct an Expert System?”
“Inference
engines”
“How does it work?”
The forward chaining mechanism
The backward chaining mechanism
Logic used in the Expert Systems
What is an expert system shell?
Applications of the expert systems
“What is explicit, what is implicit?”
“But where is
the
knowledge about
how to
resolve the problem?”
Knowledge acquisition is a modellisation
The main point is the construction of the conceptual model
“Which are the components of a conceptual model?”
“Why is the modelisation necessary?”
“What kind of problem solver to construct?”
“Why reasoning the same way the expert does?”
Need for a constructive approach
The existing constructive approaches lack the followings:
Principle of the structural correspondence
Prototyping at the knowledge level
Construction of a conceptual model
Languages with predefined modelling primitives
“I know that
I can make mistakes!”
Objective of the Mapcar language
Architecture of the Mapcar language
Examples of inference rules in Mapcar
Conceptual model
layers
¹
ZoLa sub-languages
Architecture of the ZoLa (properties)
Architecture of the ZoLa (functionality)
Examples of the ZoLa constructions
L2 profil and L2 operation
(type manipulation)
Notions of the model (conceptual model of the domain)
High level actions of DSTM (conceptual model of the reasoning)
Architecture with an algorithmic control
“Is this control is good enough? Can we do better?“
Architecture with an dynamic control
Architecture with an dynamic control (2)
Connexion with conceptual handles
Dictionnary to translate between the layers
Connexion layer (control algorithmic)
Different forms of reflexivity, some “definitions”
What happens if we start the reasoning?
SADE supervise the knowledge based system…