1

Artificial IntelligenceArtificial Intelligence

University Politehnica of Bucharest2008-2009

Adina Magda Floreahttp://turing.cs.pub.ro/aifils_08

2

Lecture No. 6

Prolog Language

Rule-based Systems Knowledge representation using rules RBS structure Inference cycle of a RBS Control strategy

2

3

1. Prolog R. Kowalski, A. Colmerauer - '70

Prolog control structure A goal S is true in a Prolog program (can be

satisfied or is a logical consequence of the program) iff:

1.  There is a clause C in the program;

2.  There is an instance I of the clause C such that: 2.1.   the head of I is the same with the head of S;

2.2 all the goals in I’s body has true (can be satisfied)

3

4

Prolog

Execuþie

sistem Prolog

conjuncþiide scopuri

program Prolog

indicator SUCCES/INSUCCES

instanþele variabilelor din scopuri

5

Prolog% parinte(IndividX, IndividY)% stramos(IndividX, IndividZ)parinte(vali, gelu).parinte(ada, gelu).parinte(ada, mia).parinte(gelu, lina).parinte(gelu, misu).parinte(misu, roco).

str1(X, Z) :- parinte(X, Z).str1(X, Z) :- parinte(X, Y), str1(Y, Z).

5

vali ada

gelu mia

lina mişu

roco

6

Prolog% Change the order of the rules:

str2(X, Z) :- parinte(X, Y), str2(Y, Z).

str2(X, Z) :- parinte(X, Z).

% Change the order of goals:

str3(X, Z) :- parinte(X, Z).

str3(X, Z) :- str3(X, Y), parinte(Y, Z).

% Change both the order of rules and the order of goals:

str4(X, Z) :- str4(X, Y), parinte(Y, Z).

str4(X, Z) :- parinte(X,Z).

6

7

Prolog

?- str1(ada, misu).

yes

?- str2(ada, misu).

yes

?- str3(ada, misu).

yes

7

str1(ada, misu)

parinte(ada, misu)

INSUCCES

parinte(ada, Y), str1(Y, misu)

parinte(ada, gelu)

Y=gelu

SUCCES

str1(gelu, misu)

parinte(gelu, misu)

SUCCES

str3(ada, misu)

parinte(ada, misu)

INSUCCES

str3(ada, Y), parinte(Y, misu)

parinte(ada, Y')

Y=Y'

SUCCES

parinte(gelu, misu)

SUCCES

parinte(ada, gelu)

Y'=gelu

8

Prolog?- str4(ada, misu).

?- str3(mia, roco).

8

str4(ada, misu)

str4(ada, Y), parinte(Y, misu)

str4(ada, Y'), parinte(Y', Y)

str4(ada, Y"), parinte(Y", Y)

arbore infinit...

str3(mia, roco)

parinte(mia, roco)

INSUCCES

str3(mia, Y), parinte(Y, roco)

parinte(mia, Y)

INSUCCES

str3(mia, Y'), parinte(Y', Y)

parinte(mia, Y')

INSUCCES

str3(mia, Y"), parinte(Y", Y')

parinte(mia, Y")

INSUCCES

...arbore infinit

9

2. Knowledge representation using rules Declarative representation / procedural representation

R1: if the patient has fever

and the type of the organism is gram-pozitiv

and the patient has a stiff neck

then the identity of the organism is streptococcus

R2: if the car does not start

and the lights do not turn on

then the battery is faulty

sau there is no contact of the battery

R3: if the temperature > 95o C

then open the protection valve

10

Knowledge representation using rules

( p)( o)(Temperatura (p,mare) Tip(o,gram - pozitiv) GitUscat(p))

Identitate (o,streptococ)

( m)( b)(NuPorneste(m) NuAprinde(m,far))

(Stare(b,consumata) Borne(b, fara_ contact))

3. Structure of an RBS

MEMORIEDE LUCRU

Fapte dinamice

BAZA DE CUNOSTINTE Reguli Fapte

Date de activare

MOTOR DEINFERENTA

Selectie reguliReguli selectate

Actiuni(Concluzii)

Date

Reprezentare

Control

Intrari

(date de caz)

Iesiri

(Raspunsuri)si fapte

12

4. Inference cycle of an RBS

Match Selection Execution

13

Inference cycle of an RBS - cont

Algorithm: Functioning of an RBS

1. WM Case data

2. repeat

2.1. Identify WM and KB and create the conflict set (CS)

2.2. Select a rule from the CS to apply

2.3. Apply selected rule

until there are no more applicable rules or

WM satisfies goal or

a predefined effort has been exhausted

end.

14

5. Control strategy

Selection criteria from the CS Direction of rule application

14

15

5.1 Selection criteria from the CS

Select the first applicable rule Select a rule from the CS

Preferences based on rule identitySpecificity

- the set of premises of R1 includes the set of premises of R2;

- the set of conditions of R1 are similar to that of R2, but the premises of R1 refer to constants while the premises of R2 refer to variables.

Time of last use Preferences based on matched objects

15

16

17

Selection criteria from the CS - cont

Use of meta-rules

if a rule has conditions A and B and

the rule refers {does not refer} X

{ not at all/

only in the LHS/

only in the RHS }

then the rule will be certainly useful

{ probably useful/

probably useless/

certainly useless }

Application of all rules in the CS 17

18

5.2 Direction of rule application

18

INLANTUIRE INAINTE

daca atuncidaca atunci

INLANTUIRE INAPOI

daca atuncidaca atunci

daca atunci

A B B C

A (data)

C (concluzie)

determina C B C A B

( A C, implicit)Este A adevarata? (data)

Forward chaining Backward chaining

Continut initial al memoriei de lucru: A,EStare scop: D

A,E A,E,B A,E,B,C A,E,B,C,D

(a)

R3 R1 R2

A,E

A,E,B

A,E,B,C

A,E,B,C,D

A,E,C

A,E,C,D

R4

R2

R3

R1

R2

(b)

(a) Forward chaining

19

R3

Continut initial al memoriei de lucru: A,EStare scop: D

D? C? A?, B? A?, E?

(a)

R3R1R2

D

C

A B E

R1 R4

R2

A? A? E?

E?

(b)

A

C

(b) Backward chaining

20

Algorithm: Find a value using forward chaining (a)FindValueD(A)1. if A has value

then return SUCCESS2. Build CS3. return Find(A, CS)end.Find(A, CS)1. if CS = {}

then return FAIL2. Choose a rule RCS according to a selection criterion (criteria) 3. for each premise Ij of R execute

3.1. Find the truth of Ij4. if all Ij , j=1,N are satisfied

then4.1. if A has value

then return SUCCESS4.2. Add the fact(s) refered in the RHS of R to WM

5. CS CS - R6. return Find(A, CS)end.

Algoritm: Find a value using backward chaining (b) Detvalue(A)1. Build the set if rules which refer A in RHS, CS2. if CS = {}

then2.1. Ask the user the value of A2.2. if the user answers A then stare this value in WM else return FAIL

3. else3.1. Choose a rule according to a selection criterion (criteria)3.2. for each premise Ij, j=1,N, in LHS of R execute3.2.1. Be Aj refered by the premise Ij3.2.2. if Aj has valuethen evaluate the truth of Ij3.2.3. elsei. if Detvalue(Aj) = SUCCESSthen evaluate the truth of Ijii. else consider Ij unsatisfied

3.3. if all premises Ij, j=1,N are satisfied

then store the value A obtained from the RHS of R in WM

4. if A has value (in WM)

then return SUCCESS

5. else

5.1 CS CS – R

5.2 repeat from 2

end.

1'. if A is primary data

then

1'.1. Ask the user the value of A

1'.2. if the user knows the value of A

then - store this value in WM

- return SUCCESS

1’.3. else return FAIL

Example of RBS with forward chaining (a)

R11: if Etapa is Verifica-Decor

and exists o statuie

and does not exist soclu

then Add soclu la Obiecte-Neimpachetate …..

R17: if Etapa is Verifica-Decor

then finish Etapa Verifica-Decor

and start Etapa Obiecte-Mari

Nume Greutate Dimensiune Fragil Impachetat

Etapa :Cutie 1 :Obiecte Neimpachetate :

Statuie mediu mare nu nuTablou usor medie nu nuVaza usor mica da nuSoclu greu mare nu nu

Verifica - Decor

Statuie,Tablou,Vaza ,Soclu{ }

Example of RBS with forward chaining - contR21: if Etapa is Obiecte-Mari

and exists un obiect mare de ambalatand exists un obiect greu de ambalatand exists o cutie cu mai putin de trei obiecte marithen pune obiectul greu in cutie

R22: if Etapa is Obiecte-Mariand exists un obiect mare de ambalatand exists o cutie cu mai putin de trei obiecte mari …..then pune obiectul mare in cutie

R28: if Etapa is Obiecte-Mariand exists un obiect mare de pusthen foloseste o cutie noua

R29: if Etapa is Obiecte-Marithen finish Etapa Obiecte-Mariand start Etapa Obiecte-Medii

25

Etapa :Cutie 1 :Obiecte Neimpachetate :

Obiecte - MediiSoclu,StatuieTablou,Vaza

Example OPS5WME

(literalize student name placed_in sex_stud)

(literalize room number capacity

free sex_room occupants)

(vector-attribute occupants)

room 111 4 2 F (Maria Elena) time tag

(make room ^number 112 ^capacity 4 ^free 4

^sex_cam nil ^occupants nil)

(make student ^name Mihai ^placed_in nil ^sex-stud M)

26

(p atrib_stud_free_room

(<Unplaced_student>

(student ^name < stud_name> ^plased_in nil ^sex_stud <gender>))

(<Free_room>

(room ^number <room_no>

^capacity <capacity_max>

^free <capacity_max>))

-->

(modify < Unplaced_student > ^ plased_in < room_no >)

(modify < Free_room > ^ occupants < stud_name >

^sex_room < gender >

^ free (compute < capacity_max >-1)).

27

28

Example Jess

(deftemplate person(slot firstname)(slot name)(slot age (type INTEGER))

)

(defrule adult(person (firstname ?p_first)

(age ?p_age))(test (>= ?p_age 18)))

=>(assert (adult (firstname ?p_first)))

)

Stud/room in Jess(deftemplate student

(slot name)(slot sex)(slot placed_in)(slot special_considerations (default no))

(deftemplate room(slot number)(slot capacity (type INTEGER)(default 4))(slot sexes_are)(slot vacancies (type INTEGER))(multislot occupants))

(defrule assign-student-empty-room?unplaced_student (student (name ?stud_name)

(placed_in nil) (sex ?gender))

?empty_room (room (number ?room_no) (capacity ?room_cap) (vacancies ?max_size))

(test (= ?room_cap ?max_size))=>

(modify ?unplaced_student (placed_in ?room_no))(modify ?empty_room (occupants ?stud_name) (sexes_are ?gender)

(vacancies (-- ?max_size)))

30

5. RBS strategies in OPS5 family

Recognize-act cycle: Match Select Act

WME = working memory element Identified by a "time tag"

Instantiation: set of WME which satisfy a rule Conflict set Conflict resolution

30

31

Recognize-act cycle

Match Application of rule – condition - WMEs of WM

satisfy LHS 2 aspects:

match intra-element match inter-element

(defrule teenager(person (firstName ?name) (age ?age))

=> (printout t ?name " is " ?age " years old." crlf))

Recognize-act cycle match inter-element(defrule assign-private-room

(student (name ?stud_name)

(placed_in nil)

(special_consideration yes))

(room (number ?room_no)

(capacity 1)

(vacancies 1)

=> …

WMEs41 (student name Mary sex F placed_in nil

special_consideration yes)52 (student name Peter sex M placed_in nil

special_consideration yes)9 (room number 221 capacity 1 vacancies 1)12 (room number 346 capacity 2 vacancies 1)17 (room number 761 capacity 1 vacancies 1)

Instantiations (WM)assign-private-room 41 9assign-private-room 41 17assign-private-room 52 9assign-private-room 52 17

33

Conflict resolution

Strategies Refraction = o aceeasi instantiere nu este

executata de mai multe ori (2 instantieri sunt la fel daca au acelasi nume de regula si aceleasi time tags, in aceeasi ordine)

Time of use = Se prefera instantierile care au identificat cu WMEs cu cele mai recente time tags (sau invers)

Specificity = Au prioritate instantierile cu LHS specifice = nr de valori testate in LHS

teste asupra: nume clasa, predicat cu 1 arg constanta, predicat cu un operator variabila legata

34

Conflict resolution

Strategia LEX Elimina din MC instantierile care au fost deja

executate Ordoneaza instantierile pe baza momentului

utilizarii Daca mai multe instantieri au aceleasi time

tags, utilizeaza specificitate Daca mai multe instantieri au aceeasi

specificitate alege arbitrar Strategia MEA – aceeasi dar utilizeaza primul

time tag

35

Efficiency

Match – cam 80% din timpul ciclului recunoastere-actiune Fiecare WME este comparat cu fiecare conditie din fiecare

regula O(R*FP), R – nr de reguli, F – nr WME, P nr mediu de

conditii in LHS regula RETE match algorithm (1982 Forgy)

Salveaza starea de match intre 2 cicluri recunoastere-actiune

La crearea unui WME, acesta este comparat cu toate elementele conditie din program si este memorat impreuna cu fiecare element conditie cu care a identificat =>

Numai schimbarile incrementale din WM sunt identificate in fiecare ciclu

O(R*F*P) aprox

Rule compilation Se compileaza conditiile regulilor intr-o retea de noduri de

test Un test este un predicat cu o constanta sau variabila legata

sau o disjunctie Procesul de match are loc numai la adaugarea sau la

eliminarea unui WME (modify este retract urmat de assert)(gate (type or) (value true))(gate (type or) (value false))

Node sharing1 = gate

type = or

value = true value = false

37

Rule compilation

Un pointer la noul WME este trecut prin retea, incepand de la nodul radacina

Fiecare nod actioneaza ca un switch Cand un nod primeste un pointer la un WME, testeaza

WME asociat. Daca testul reuseste, nodul se deschide si WME trece mai departe

Altfel nu se intampla nimic Daca un WME va trece prin retea, va fi combinat cu

alte WME care trec pentru a forma o instantiere in MC Pointerii la WME sunt trimisi prin reteaua RETE ca

tokens = pointer + stare (assert sau retract)

38

Types of node in RETE network

Nod cu 1 intrare = test intra-element realizat de noduri cu 1 intrare fiecare nod efectueaza un test corespunzator unei conditii testarea aceleiasi variabile – tot noduri cu 1 intrare

1 = gate

type = or

value = true value = false

39

RBS

Foarte multe Cele mai influente

OPS5 ART CLIPS Jess Familia Web Rule languages

In special RuleML si SWRL Interoperabilitatea regulilor

40

RuleML

RuleML Initiative - August 2000 Pacific Rim International Conference on Artificial Intelligence.

RuleML Initiative dezvolta limbaje bazate pe reguli deschise XML/RDF

Aceasta permite schimbul de reguli intre diferite sisteme, inclusiv componete software distribuite pe Web si sisteme client-server eterogene

Limbajul RuleML – sintaxa XML pentru reprezentarea cunostintelor sub forma de reguli

Integrarea cu ontologii: sistemul de reguli trebuie sa deriveze/utilizeze cunostinte din ontologie

41

RuleML

RuleML – se bazeaza pe Datalog Datalog = limbaj de interogare si reguli pentru baze de

date deductive Subset al Prolog cu restrictii:

argumente ne-functionale (constante sau variabile) limitari in nivelul de apeluri recursive variabilele din concluzie trebuie sa apara in predicate

ne-negate din ipoteza Hornlog – Datalog extins cu variabile functionale (termeni

generali)

42

RuleML

Reguli reactive = Observa/verifica anumite evenimente/conditii si executa o actiune – numai forward

Constrangeri de integritate = reguli speciale care semnaleaza inconistente cand se indeplinesc anumite conditii – numai forward

Reguli de inferenta (derivare) = reguli reactive speciale cu actiuni care adauga o concluzie daca conditiile (premisele sunt adevarate) - Se pot aplica atat forward cat si backward

Fapte = reguli de inferenta particulare

Regui reactive

Constangeri deintegritate

Reguli dederivare

Fapte

RuleML Reguli reactive<rule> <_body> <and> prem1 ... premN </and> </_body>

<_head> action </_head></rule>

Constrangeri de integritate<ic> <_head> inconsistency </_head>

<_body> <and> prem1 ... premN </and> </_body></ic>

implementate ca

<rule> <_body> <and> prem1 ... premN </and> </_body> <_head> <signal> inconsistency </signal> </_head>

</rule>

RuleML Reguli de inferenta/derivare

<imp> <_head> conc </_head> <_body> <and> prem1 ... premN </and> </_body>

</imp >

implementate prin<rule> <_body> <and> prem1 ... premN </and> </_body>

<_head> <assert> conc </assert> </_head></rule>

Fapte<atom> <_head> conc </_head> </atom> implementate prin<imp> <_head> conc </_head>

<_body> <and> </and> </_body> </imp>

RuleMLFapte

<atom> <rel>spending</rel> <ind>Peter Miller</ind> <ind>min 5000 euro</ind> <ind>previous year</ind>

</atom>

spending(petterMiller, min5000euro, previousYear).

Reguli de inferenta/derivare<imp> <head> <atom>

<rel>discount</rel> <var>customer</var>

<var>product</var> <ind>7.5 percent</ind> </atom>

</head> <body> <and> <atom>

<rel>premium</rel> <var>customer</var> </atom> <atom> <rel>luxury</rel> <var>product</var> </atom>

</and> </body></imp>

discount(Customer, Product, 7.5_percent):-premium(Customer), luxury(Product).

Example of RBS with backward chaining (b)R1: if X has par

then X is mamalR2: if X feeds puii with milk

then X is mamalR3: if X is mamal

and X has pointed teethand X has falcithen X is carnivor

R4: if X is carnivorand X is brownand X has petethen X is cheetah

R5: if X is carnivorand X is brownand X has stripsthen X is tiger

47

What is X?