Implementing knowledge bases 1. Concepts in Knowledge representation methods • Some of the techniques applied include o Relational Databases o Production Rules o Logic Representation o Frames o Case-bases • Graphic Representations include: o Semantic Networks o Decision Trees o

2. Production Rules Production Rules are program codes that follow the format IF…THEN…ELSE Eg IF international Conflict Begins AND it is in Middle East THEN oil price goes up.

✓ Symbolically: ✓ IF A AND B THEN C
Example
Consider the following information given by consumers surveying bureau of Kenya
Let the rules and symbols be as follows: ✓ “if one leaves near Muthaiga road (A)and has a plump salary(B) then he is likely to join muthaiga golf club(C). ✓ One can say he leaves near Muthaiga road(A) if he leaves less than 15 kilometers off the road.(D) ✓ The survey also reveals that you have to have a plump income (B) and be a who-is-who in Nairobi(E) so that you can join an exclusive club(F). ✓ If you can join an exclusive club(F) then you are likely to meet politicians(G). ✓ One who meets politicians (G)is said to be well connected(H). ✓ If one can join exclusive (F) and is a who-is-who in Nairobi (E) then he is well connected.(H) ✓ If well connected then (H)one can join Muthaiga golf club (C). ✓ If one can join Muthaiga golf club (C) then he is likely to meet the president (I).”
Thus using symbolic representation, the production-rule knowledge base containing the above information would be as follows. IF A AND B THEN C IF D THEN A IF B AND E THEN F IF F THEN G IF G THEN H IF F AND E THEN H IF H THEN C IF C THEN I

3. LOGIC: Propositional and predicate calculus
a) Propositional Logic. ✓ For a computer to perform reasoning using logic, some method must be used to convert statements and the reasoning process into a form suitable for manipulation. The result is ‘symbolic logic’. ✓ Symbolic Logic is a system of rules and procedures that permit the drawing of inferences from various premises using a variety of logical techniques. ✓ There are two types of computational logic: ✓ Propositional logic ✓ Predicate logic. ✓ NB: Term calculus is sometimes used instead of logic and means ‘system of computing’

Input; premises, facts, propositions etc Output; Inferences, Conclus

Fig. 2.3.1 : Using Logic to Reason.

Representing Simple facts in Propositional Logic

✓ A proposition is a statement that can be evaluated to be true or false. ✓ The statement can be used to derive new propositions of conclusions. ✓ It is usual to use symbols, such as letters of the alphabet or abbreviations, to represent various propositions, or conclusions. ✓ One can use logical connectives such as AND, OR, NOT, IMPLIES, EQUIVALENT. ✓ NB: Propositional Logic deals with complete statements and whether they are true or false. ✓ Example to illustrate propositional logic ✓ Its Raining ✓ Its Sunny ✓ Its Windy

b) Predicate Logic.
Statement, Components, Facts and Variables. ✓ In predicate logic, one can break a statement into component parts. ✓ These parts are either: ✓ Object ✓ Characteristic ✓ Relationship ✓ Some assertion about the object. ✓ In addition, predicate calculus lets us use variables in a symbolic logic statement Example 1 Jane is naughty. She combs her hair in class is (jane, naughty). combs (jane, hair_class).

Example 2 Mwangi travels by matatu, bus, bicycle. travels (mwangi, matatu) travels (mwangi, bus). travels (mwangi, bicycle). travels (mwangi, X). X= {matatu, bus, bicycle}

Note: In Prolog representation, upper case letters represent variables.

Rules representation in Predicate Logic

✓ We can also represent Rules: ✓ Eg. If you show symptoms of sneezing, fever and shivering then you have a disease called flu.

symptom_sneeze ( symptom_fever ( symptom_shiver ( disease (flu)

In Prolog

is(symptom, sneeze) , is(symptom ,fever) , is(symptom, shiver) :- is(disease , flu).

Exercise: Represent the following statements:

1. All prehistoric tribes of central Kenya died when mount Kenya erupted as a volcano in 600 BC. 2. If someone dies, then he is dead at all later times. 3. I was born in 1990. 4. If born before 1970, then the candidate probably sat for ‘A’ levels

Formal Approach- well formed formula (wff). ✓ The representation of real world facts as logical propositions using the standard logic symbols results in a series of clauses known as ‘well-formed formulas’ i.e. wff’s ✓ We occasionally use standard symbols to represent logical connectives. Thus: ( IMPLIES ➢ NOT ← OR ← AND ← FOR ALL ← THERE EXISTS Example 1: Represent the following statements as propositions using ‘well-formed formula’ (wff): People only try to assassinate rulers they are not loyal to. Ochuka tried to assassinate Moi. ??Was Ochuka loyal to Moi?? Ans:

(x: (y: person(x) ( ruler(y) ( (loyalto(x,y) ( tryassassinate(x,y). tryassassinate ( ochuka, moi). ?-: loyalto (ochuka, moi ). Note ✓ The above statements seem as if they can yield a response (‘NO’) but it may not because we don’t know if Ochuka is a person and Moi was a ruler. ✓ Therefore, we need to explicitly state the two facts for evaluation to be successful. ✓ Example 2 ‘ All men are people’

(X: Man(X) ( Person(X)

Example 3 ‘If it is not raining, then it is sunny’

(RAINING ( SUNNY Is converse true? RAINING (( SUNNY

Example 4 All Kenyans are loyal to Kibaki or dislike him

( x : Kenyan(x)( loyalto(x, Kibaki)( dislike(x, Kibaki).

Example 5: Every one is loyal to someone

(x : (y : loyalto (x,y)

Further examples and also incorporating prolog implementation
Example 1 IF there is a lot of oil demand in China THEN global oil prices rise.

(china_oil_demand, a_lot) ( will(global_oil_price, rise).

In Prolog

is (china_oil_demand, a_lot) :- will(global_oil_price, rise).
Example 2 Either Kenya, Uganda or Tanzania will join the U.N. security council.

(Kenya ( uganda ( tanzania)( join_council

In Prolog.

will( join_council , kenya) ; will ( join_council, uganda) ; will( join_cou ncil, tanzania).
Example 3 For Inflation to occur there must be bad export business, poor economic management and weak currency.

bad_export ( poor_economic_management ( weak_currency ( inflation

In Prolog

(bad_export , poor_economic_management , weak_currency) :- inflation

Exercise 1. Write the meaning of the following statement (x : (t1 : (t2 : mortal (x) ( born(x , t1) ( greater_than( t2 - t1 , 150) ( dead(x , t2). 2. Write the following rule using production rule, standard logic symbols and then as prolog statement.
“Someone is sister of another if that person is a female and both have same father and same mother”

3. Makori is a man. He is a Kisii. All Kisiis are Kenyans. He must be a Kenyan ANS (x : (y: kisii(x) ( kenyan(y) ( man(makori) ( is(makori, x) ( is(x, y) (is (makori, y)

a) Applications to Logic Design and formal system specifications.

✓ The above symbolic representation is normally used to represent simple facts, complex facts (have to be broken down), rules, relationships (predicates) and other types of knowledge. ✓ It is useful in AI because we can use the representation and then perform computations also known as inferencing (or reasoning). ✓ Some popular reasoning strategies include: ✓ Modus Ponens ✓ Matching ✓ Forward and backward chaining ✓ Resolution ✓ Natural deduction ✓ We can write applications that use the above techniques to solve practical problems. ✓ Expert Systems can be programmed easily using above techniques. ✓ Systems that require matching eg pattern matching. ✓ Systems that use ‘look-ups’ eg Language Translators, Dictionaries, Games, vision systems etc

4. Frames
The method is commonly known as Frame Representation Scheme ✓ A frame is a collection of attributes, usually called slots. ✓ The slots are associated with facets. Facets contain values, constraints in to the values, procedures etc ✓ We usually build a frame system from a collection of frames that are connected to each other by virtue of the fact that the value of an attribute of one frame may be another frame. ✓ Set theory provides a good basis for understanding frame systems. ✓ Each frame may be implemented as a class (or a set) in an object oriented environment.

Fig. 2.3.2 Illustrating a Frame System Frames are represented in hierarchies and show inheritance 5. Semantic Networks Semantic networks are graphic representations that show objects their relationships or meanings. Example Consider the following information “ In the animal world, carnivores such as lions and leopards hunt and eat herbivores such as zebras and wild beast. Thus some animals are hunters and some are hunted, and generally speaking carnivores hunt herbivores. However there are many exceptions to this rule. Elephants, which are herbivores, are rarely hunted because of their size. Nor are bears, which are not normally meat eaters and therefore basically herbivores. Bears will catch and eat other animals when opportunity arises.” i) Draw a well-labeled semantic network to represent the above information.

All connections are named isa.
Note: Semantic networks can easily be converted to prolog databases by converting the relationships into predicates. ii) Write a knowledge base in PROLOG that can be used to represent the above information.
Ans
Isa(carnivore,lion). Isa(carnivore,leopard). Isa(carnivore,bear). Isa(hunter,lion). Isa(hunter,leopard). Isa(hunter,bear). Isa(herbivore,elephant). Isa(herbivore,zebra). Isa(herbivore,wildebeeste). Isa(herbivore,bear). Isa(hunted,zebra). Isa(hunted,wildebeeste).
Example 2
Consider the following information “ Parents in the U.K. are faced with a variety of schools. There are primary schools for young children and secondary schools for older children. Younger children can go to public primary schools or private schools. Older ones mostly go to comprehensive schools, but there are also grammar schools for clever pupils, city technology colleges for clever ones, and fee paying public schools. Grammar and public schools offer a more academic education.”

From the above information i) Draw a well labeled semantic network

Goto is_a Is_a goto

Is_a

Is_a Goto Is_a Is_a

Goto is_a

Is_a is_a

-----------------------
Logical Process

FRAME:FACULTIES

FRAME: UNIVERSITY

SCI

FACULTIES

POPULATION: 6,000

Engineering

FRAME: SCI

ADIMISS. CRITERIA

MEDICINE

Admission:

Lecturers: 22

FRAME:ADMISSION

GRADES

PROCESS

Lion

Zebra

Hunted

Hunter

Carnivore

Bear

Herbivore

Leopard

Wildebeest

Elephant

Comprehensive sch

Young_ children

Older children

Public pri school

Grammar sch

Primary school

Sec. school

Private school

CTC

Clever

Academic pupils

Fee paying

Public school