Experts System

Gerstner Laboratory for Intelligent Decision Making and Control

Expert Systems I Michal Pěchouček

Gerstner Laboratory for Intelligent Decision Making and Control

Expert System Functionality • replace human expert decision making when not available • assist human expert when integrating various decisions • provides an ES user with – an appropriate hypothesis – methodology for knowledge storage and reuse • border field to Knowledge Based Systems, Knowledge Management • knowledge intensive × connectionist • expert system – software systems simulating expert-like decision making while keeping knowledge separate from the reasoning mechanism Gerstner Laboratory for Intelligent Decision Making and Control

Expert Systems Classification • Unlike classical problem solver (GPS, Theorist) Expert Systems are weak, less general, very case specific • Exert systems classification: – Monitoring – Interpretation – Repair & – Prediction Debugging – Diagnostic – Instruction – Design & Configuration – Control – Planning

Gerstner Laboratory for Intelligent Decision Making and Control

Underlying Philosophy •

knowledge representation

– – – –

•

production rules logic semantic networks frames, scripts, objects

reasoning mechanism

– knowledge-oriented reasoning – model-based reasoning – case-based reasonig

Gerstner Laboratory for Intelligent Decision Making and Control

Expert System Architecture

knowledge base

knowledge base editor

inference engine

user

world model

preceptors

explanation subsystem

Gerstner Laboratory for Intelligent Decision Making and Control

Rule-Based System • •

knowledge in the form of (production) rules reasoning algorithm: (i) (ii) (iii) (iv)

•

• •

if condition then effect

FR ← detect(WM) R ← select(FR) WM ← apply R goto (i)

conflicts in FR: – first, last recently used, minimal WM change, priorities incomplete WM – querying ES (art of logical and sensible querying) Gerstner Laboratory examples – CLIPS (OPS/5), Prolog for Intelligent Decision Making and Control

Rule-Based System Example here → fine not here → absent absent and not seen → at home absent and seen → in the building in the building → fine at home and not holiday → sick here and holiday → sick

? here → no

not here, in the building → fine not here, not holiday → sick

? here → yes

? seen → no ? holiday → no sick ? here → yes fine

? holiday → yes sick Gerstner Laboratory for Intelligent Decision Making and Control

Data-driven × Goal-driven here

seen

holiday data driven

absent

home

building

goal driven fine

sick

Gerstner Laboratory for Intelligent Decision Making and Control

Data-driven × Goal-driven • goal driven (backward chaining) ~ blood diagnostic, theorem proving – limited number of goal hypothesis – data shall be acquired, complicated data about the object – less operators to start with at the goal rather than at the data • data driven (forward chaining) ~ configuration, interpretation, – reasonable set of input data – data are given at the initial state – huge set of possible hypothesis Gerstner Laboratory for Intelligent Decision Making and Control

Knowledge Representation in ES •

Knowledge Models – rules, frames, logic, networks – first generation expert systems • Deep Knowledge Models – describes complete systems causality – second generation expert systems • Case Knowledge Models – specifies precedent in past decision making Shallow

Gerstner Laboratory for Intelligent Decision Making and Control

Model Based Reasoning • Sometimes it is either impossible or imprecise to describe the domain in terms of rules … • Here we use a predictive computational model of the domain object in order to represent more theoretical deep knowledge model • Model is based either on – quantitative reasoning (differential equations, …) – qualitative reasoning (emphasizes some properties while ignoring other) • Very much used for model diagnosis and intelligent tutoring Gerstner Laboratory for Intelligent Decision Making and Control

Qualitative Reasoning •

Qualitative Reasoning

•

QR Techniques:

is based on symbolic computation aimed at modeling of behavior of physical systems – commonsense inference mechanisms – partial, incomplete or uncertain information – simple, tractable computation – declarative knowledge – Constrain based – Qualitative Simulation QSIM – Component based – Envision – Process based – QPT (Qualitative Process Theory) Gerstner Laboratory for Intelligent Decision Making and Control

QSIM – A Constraints Based Approach •

Qualitative system is described by parameters, domains constraints (relations among parameters)

•

Qualitative simulation

and

is thus only breath-first-search in the space of possible combination of values of the parameters • Qualitative behaviour is thus a path in the tree from the initial state to some leaf state • The structure of the system model is given in the form of qualitative equation consisting of constraints: – arithmetic – add(A,B,C),mult(A,B,C) – derivative – der(height, velocity) – monotonicity – M+(wrinkle,age) M(hunger,consumption)

Gerstner Laboratory for Intelligent Decision Making and Control

QSIM – A Constraints Based Approach •

of each parameter is a couple: {value,direction} where value can be either an or landmark value and direction may be inc (increasing), dec (increasing) or std (steady) • Qualitative Reasoning Procedure: Qualitative State

(i) (ii) (iii) (iv)

•

interval

wm ← initial state succ ← find-successors of first(wm) succ ← filter(succ) wm ← wm – first(wm) + succ

Filtering:

pairwise consistency, redundancy, cycles, termination condition, logical direction of change, qualitative magnitude change Gerstner Laboratory for Intelligent Decision Making and Control

QSIM – A Pendulum Example • system description: der(v,a) and der(s,v) • domains: a = {min,〈min,0〉,0,〈0,max〉,max} v = {min,〈min,0〉,0,〈0,max〉,max} s = {0,〈0,max〉,max}

v s

s

0,std

+,inc

+,inc

+,inc

max,st d

v

0,std

+,inc

max,st d

+,dec

0,std

a

max,st d

+,dec

0,std

-,dec

min,std

s

max,st d

+,dec

+,dec

+,dec

0,std

v

0,std

-,dec

min,std

-,inc

0,std

a

min,std

-,inc

0,std

+,inc

min,std

a

Gerstner Laboratory for Intelligent Decision Making and Control

Case Based Reasoning • part of the machine learning lecture • Algorithms: – problem attributes description – retrieval of previous case – solution modification – testing new solution – repairing failure or inclusion into the plan library • Utilized widely in law domain (Judge)

Gerstner Laboratory for Intelligent Decision Making and Control

Knowledge Evolution •

- result of application of the knowledge extraction process on the set E ∪ S. • Weak Update - relevant bits of the inference knowledge-base re-computation Strong

Update

's t r o n g ' u p d a t e

s tro n g u p d a te

I L P

in f e r e n c e r u le s

e x c e p tio n

w e a k u p d a te

F ilt e r s

d e c is io n g ra p h

EBG

w e a k u p d a te

Gerstner Laboratory for Intelligent Decision Making and Control