Inteligena Artificial Cursuri 3-4

Noiuni de prelucrarea limbajului natural

Tehnologiile limbajului
Tehnologiile moderne ale limbajului i cercetarea lingvistic ne ajut n activiti zilnice:
cnd efectum cutri pe internet, cnd traducem pagini web, cnd folosim facilitile de corectare a ortografiei i gramaticii oferite de un procesor de texte, cnd dm comenzi vocale aparaturii din main sau telefonului mobil, cnd primim recomandri n librrii online, cnd urmm indicaiile unei aplicaii de navigare prin GPS.

mprumutat de la Diana Trandab

Fraze
Solicitat s comenteze un editorial recent al lui Dinu Patriciu, n care acesta preciza c nu crede n social-liberalism i s aprecieze dac, astfel, a dat o lovitur de imagine USL, Antonescu a spus c nu tie dac Patriciu s-a referit la USL. | El a precizat c USL este o uniune social-liberal pentru c reunete un partid social-democrat cu un partid naional liberal. |

Propoziii
Solicitat s comenteze un editorial recent al lui Dinu Patriciu,| n care acesta preciza |c nu crede n social-liberalism| i s aprecieze |dac, astfel, a dat o lovitur de imagine USL,| Antonescu a spus |c nu tie |dac Patriciu s-a referit la USL.|

Uniti lexicale
Solicitat s comenteze un editorial recent al lui Dinu Patriciu , n care acesta preciza c nu crede n social-liberalism i s aprecieze dac , astfel , a dat o lovitur de imagine USL , Antonescu a spus c nu tie dac Patriciu s - a referit la USL .

Grupuri nominale
Solicitat s comenteze [un editorial recent al lui [Dinu Patriciu]], n [care] [acesta] preciza c nu crede n [social-liberalism] i s aprecieze dac, astfel, a dat [o lovitur de [imagine]] [USL], [Antonescu] a spus c nu tie dac [Patriciu] s-a referit la [USL].

Entiti de tip nume

Solicitat s comenteze un editorial recent al lui Dinu Patriciu, n care acesta preciza c nu crede n social-liberalism i s aprecieze dac, astfel, a dat o lovitur de imagine USL, Antonescu a spus c nu tie dac Patriciu s-a referit la USL.
persoan perso an

organizai persoan e

Roluri semantice
Carla bought the computer from Sally for $100.
buy er

goo goo ds ds

selle r

pric pric e e

Sally sold the computer to Carla for $100.

Identificarea rolurilor semantice

Cine, ce, unde, cnd, de ce, cum face o aciune Grupul NLP v invit cu drag vineri, 8 martie, la et 5 Romtelecom pentru a v mulumi pentru suportul acordat.

Lanuri corefereniale
Winston was just taking his place in one of the middle rows when two people whom he knew by sight, but

had never spoken to, came unexpectedly into the room.

One of them was a girl whom he often passed in the corridors. He did not know her name, but he knew that she worked in the Fiction Department.

10

Lanuri corefereniale
Winston was just taking his place in one of the middle
rows when two people whom he knew by sight, but had never spoken to, came unexpectedly into the room. One of them was a girl whom he often passed in the corridors. He did not know her name, but he knew that she worked in the Fiction Department.

11

Lanuri corefereniale
Winston was just taking his place in one of the middle
rows when two people whom he knew by sight, but had never spoken to, came unexpectedly into the room. One of them was a girl whom he often passed in the corridors. He did not know her name, but he knew that she worked in the Fiction Department

12

Lanuri corefereniale
Winston was just taking his place in one of the middle
rows when two people whom he knew by sight, but had never spoken to, came unexpectedly into the room. One of them was a girl whom he often passed in the corridors. He did not know her name, but he knew that she worked in the Fiction Department.

13

Structura retoric a discursului

1. Cnd a ajuns Apolo n muntele Parnas, 2. dihania uria s-a avntat spre dnsul, 3. dornic s-l ucid. 4. Dar zeul i-a ntins arcul. 5. A tras prima sageat. 6. Erau doar patru zile de cnd vzuse lumea, 7. i ntia lui sgeat a i nimerit monstrul.

CONTRAST

CIRCUMSTANCE

SEQUENCE

SEQUENCE

1
PURPOSE

CONCESSION

Entiti i relaii dintre ele

Referenialitate coref IsA, class-of part-of, has-as-part member-of, has-asmember subgroup-of, hasas-subgroup name-of, has-name X-interpret Rudenie parent-of, child-of sibling spouse-of grandparentof/grandchild-of uncle-of/nephewof unknown Afeciune love/hate worship friend-of, enemyof Sociale superior-of, inferior-of colleague-of

Poli i declanatori
Referenialitate: coref
Ion a ntlnit-o pe Maria pe prtia de ski. El fcea ski iar ea plac.

antecede nt

anafor

Poli i declanatori
Referenialitate: coref
Ion a ntlnit-o pe Maria pe prtia de ski. El fcea ski iar ea plac.

antecede nt

anafor

Poli i declanatori
Referenialitate: coref
Ion a ntlnit-o pe Maria pe prtia de ski. El fcea ski iar ea plac.

antecede nt

anafor

Poli i declanatori
Rudenie: parent-of
mama Mariei
declanat or
destinai e surs

Poli i declanatori
Afeciune: love
Ion o iubete pe Maria.

declanat or surs

destinai e

Poli i declanatori
Social: superior-of
Ion s-a ntlnit cu eful lui la o bere.

declanat or surs

destinai e

Cursuri 5-6
Proiectul: MappingBooks Let me jump in the book!

Relations
Anaphoric Non-anaphoric:
kinship affection social

SpeD Cluj-Napoca, 15-17

Anaphoric (referential) relations

coref member-of, has-as-member (inverse) isa, class-of (inverse) part-of, has-as-part (inverse) subgroup-of, has-as-subgroup (inverse) has-name, name-of (inverse)

SpeD Cluj-Napoca, 15-17

Kinship relations
parent-of, child-of (inverse) grandparent-of, grandchild-of (inverse) sibling (symmetrical) ant-uncle-of, nephew-of (inverse relations) cousin-of (symmetrical) spouse-of (symmetrical unknown
SpeD Cluj-Napoca, 15-17

Affective relations
friend-of, enemy-of (inverse) love, hate (inverse) has-fear-of worship

SpeD Cluj-Napoca, 15-17

Social relations
inferior-of, superior-of (inverse) colleague-with

SpeD Cluj-Napoca, 15-17

Poles and directionality

In case of referential:
from anaphor towards antecedent

In case of non-referential:
from source towards destination

How to identify poles

Imbricated:
the anaphor/source is larger than the antecedent/destination

Non-imbricated:
referential:
the anaphor is to the right of the antecedent

non-referential:
from source to destination, as the trigger reads

Poles and triggers

Referentiality: coref
John met Maria on the ski slope. He raced her.

antecede nt

anafor

SpeD Cluj-Napoca, 15-17

Poles and triggers

Referentiality: coref
John met Maria on the ski slope. He raced her .

antecede nt

anafor

SpeD Cluj-Napoca, 15-17

Poles and trigger

Kinship: parent-of
their father
source destinati on

trigge r

SpeD Cluj-Napoca, 15-17

Poles and trigger

Social: inferior-of Cesar s principal courtiers

destinati on

trigge r

source

SpeD Cluj-Napoca, 15-17

Poles and trigger

Affective: worship

Lygia dropped on her knees to implore someone else .

trigge r

source

destinati on

SpeD Cluj-Napoca, 15-17

Entities
Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father , a man of consular dignity from the time of Tiberius .
SpeD Cluj-Napoca, 15-17

Referential relations: coref

Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father , a man of consular dignity from the time of Tiberius .
SpeD Cluj-Napoca, 15-17

Referential relations: coref

Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father , a man of consular dignity from the time of Tiberius .
SpeD Cluj-Napoca, 15-17

Referential relations: coref

Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father , a man of consular dignity from the time of Tiberius .
SpeD Cluj-Napoca, 15-17

Referential relations: class-of

Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father , a man of consular dignity from the time of Tiberius .
SpeD Cluj-Napoca, 15-17

Kinship relations: sibling

Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father , a man of consular dignity from the time of Tiberius .
SpeD Cluj-Napoca, 15-17

Kinship relations: child-of

Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father , a man of consular dignity from the time of Tiberius .
SpeD Cluj-Napoca, 15-17

Kinship relations: parent-of

Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father , a man of consular dignity from the time of Tiberius .
SpeD Cluj-Napoca, 15-17

Kinship relations: spouse-of

Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father , a man of consular dignity from the time of Tiberius .
SpeD Cluj-Napoca, 15-17

Social relations: inferior-of

Petroniu

Vinicius was
the son of his oldest sister , who years before had married his father ,

a man of consular dignity from the time of Tiberius .

SpeD Cluj-Napoca, 15-17

Curs 7
Probleme de IA i rezolvarea lor

45

Cele 5 cerine n modelarea unei probleme de IA

Difereniaz problema general de instanele ei Recunoate o stare i apreciaz dimensiunea spaiului strilor Gsete cea mai adecvat reprezentare a strilor Reprezint tranziiile dintre stri Alege o strategie de control
46

Problem, instan de problem

8-puzzle
formulat ca o instan de problem

Misionarii i canibalii
formulat ca o instan de problem

Generarea frazelor
formulat ca o problem

Maimua i banana
formulat ca o instan de problem

Alte exemple:
jocul de ah condusul mainii...
47

Un exemplu de instan de problem

Generarea limbajului:
G1 = {N1, T1, PROP1, P1}, n care: N1 = {PROP, GN, GV, S, V} o mulime de neterminali cu semnificaiile: propoziie, grup nominal, grup verbal, substantiv i verb; T1 = {pisica, oarecele, prinde} o mulime de cuvinte; PROP1 simbolul start al gramaticii, alegerea lui semnific c ceea ce se dorete s se obin reprezint propoziii ale acestui mini-limbaj; P1 = {PROP := GN GV, GN := S, GV := V GN, S := pisica, S := oarecele, V := prinde}
48

Cutarea soluiei
Algoritmi i euristici de cutare n spaiul strilor
Strategii irevocabile
ascensional (hill-climbing)

Strategii tentative
ascensional cu revenire (backtracking)

Strategii exhaustive (brute-force)

genereaz-i-testeaz nti-n-adncime (depth-first) nti-n-lrgime (breadth-first) cel mai bun nti (best-first)

49

Curs 8
Jocuri

Regulile de joc
Doi juctori: MAX i MIN Fiecare are ca obiectiv ctigarea jocului Doar unul poate ctiga n modelarea iniial nu intervine ansa
dar ea poate fi simulat

Exemple:
ah checkers tic-tac-toe ...

Jocul tic-tac-toe
MAX joac cu X-uri MIN joac cu O-uri

Jocul tic-tac-toe
MAX

Remiz!

Jocul tic-tac-toe
MAX ctig

Reprezentarea ca o problem de IA
1. Problem versus instan 2. Spaiul strilor:
o stare: poziia pe tabla a semnelor ntre dou mutri dimensiunea spaiului: < 9+8.9+7.8.9++9! < 9.9!

3. Reprezentarea unei stri:

o matrice 3x3

4. Reprezentarea unei tranziii

algoritmic (n abordarea de fa)

5. Cum controlm evoluia jocului?

metoda MIN-MAX metoda ALPHA-BETA

Arborele de joc
MA X
o clas de simetrie

Arborele de joc
MA X MIN

Arborele de joc
MA X MIN

Arborele de joc
MA X MIN

Arborele de joc
MA X MIN

Valoarea unei stri

Ctig pentru MAX: +

Valoarea unei stri

Ctig pentru MIN: -

Evaluarea unei stri

O stare este mai bun dac deschide mai multe posibiliti de ctig pn la sfritul jocului.

Un exemplu de funcie de evaluare: valoarea strii este diferena dintre numrul de linii pe care le mai poate completa MAX i cele pe care le mai poate completa MIN.

8 7 6 5 4 3 2 1

Evaluarea unei stri

O stare este mai bun dac deschide mai multe posibiliti de ctig pn la sfritul jocului.

Un exemplu de funcie de evaluare: valoarea strii este diferena dintre numrul de linii pe care le mai poate completa MAX i cele pe care le mai poate completa MIN.

8 - 5 4 = 3 3 2 1

Evaluarea unei stri

O stare este mai bun dac deschide mai multe posibiliti de ctig pn la sfritul jocului.

Un exemplu de funcie de evaluare:

valoarea strii este diferena dintre numrul de linii pe care le mai poate completa MAX i cele pe care le mai poate completa MIN.

Liniile fr nici un semn pot fi luate de ambii juctori...

3 0 = 3

Evaluarea unei stri

2 - 1 3 = -1 2

Evaluarea: de jos n sus

MA X
1

MIN
0 -1 1 0

Evaluarea: de jos n sus

MAX alege mutarea cea mai bun pentru el
-1 -2 1

O dezvoltare a spaiului de joc pe o adncime de 2 duce la concluzia c juctorul care joac primul are o ans de ctig n plus dac ocup centrul
1

MA X MIN
1 2

0 -1 1 0

-1

MAX gndete: MIN alege mutarea cea mai bun pentru el = cea mai proast pentru mine

0 -2 -1 0

Metoda MIN-MAX
function min-max(state, player, depth) begin if (depth = 0) then return score(state); val = worst(player); while (mai sunt stri de generat) begin generez o stare -> s; val <- back-up-compare(val, min-max(s, not(player), depth-1), player); // urmtoarea micare micoreaz spaiul de cutare n cazul n care se obine poziia de ctig ntr-una // din strile generate: if (val = -worst(player)) return(val); end return(val); end function worst(player) begin if player = MAX then return - ; else return + ; end funtion back-up-compare(val1, val2, player) begin if player = MAX then return max(val1, val2); else return min(val1, val2); end

Apelul: min-max( ,MAX,2)

Evaluarea: de jos n sus

-1

1 0 -1 1

MA X
0

MIN

val=-1; player = MAX; depth=1; while (mai sunt stri de generat) begin
generez o stare -> s; ... end

Evaluarea: de jos n sus

-1

1 0 -1 1

MA X
,MIN,1)

min-max(
val=-1; player = MAX; depth=2; while (mai sunt stri de generat) begin
generez o stare -> s; val <- back-up-compare(val, min-max(s, not(player), depth-1), player); if (val = -worst(player)) return(val); end

MIN

Evaluarea: de jos n sus

-1

1 0 -1 1

s
0

MA X MIN

1
0

min-max(

,MIN,1)

val= val = worst(player); while (mai sunt stri de generat) begin generez o stare -> s; val <- back-up-compare(val, min-max(s, not(player), depth-1), player); if (val = -worst(player)) return(val); end

Evaluarea: de jos n sus

-1

1 0 -1 1

s
0

MA X MIN

-1
0

min-max( ,MAX,0)
if (depth = 0) then return score(state);

-1

Evaluarea: de jos n sus

-1

1 0 -1 1

s
0

MA X MIN

-1
0 -2 -1

min-max( val= ; player=MIN; ,MIN,1)

val <- back-up-compare(val, -1, player); if (val = -worst(player)) return(val); end

-1

Evaluarea: de jos n sus

-1

1 0 -1 1

s
0

MA X MIN

-1
0 -2 -1

min-max( val= ; player=MIN; ,MIN,1)

val <- back-up-compare(val, -1, player); if (val = -worst(player)) return(val); end

-1

-2

Evaluarea: de jos n sus

-1

1 0 -1 1

s
0

MA X MIN

-1
0 -2 -1

min-max( val= ; player=MIN; ,MIN,1)

val <- back-up-compare(val, -1, player); if (val = -worst(player)) return(val); end

-1

-2

Evaluarea: de jos n sus

-1

1 0 -1 1

s
0

MA X MIN

-1
0 -2 -1

min-max( val= ; player=MIN; ,MIN,1)

val <- back-up-compare(val, -1, player); if (val = -worst(player)) return(val); end

-1

-2

Metoda alpha-beta
-1 -1

-1 1

MA Un moment din X
dezvoltarea arborelui n care apare o situaie particular:

0 -1 1 0

MIN
Generarea poate fi oprit!
Ea nu mai poate influena valoarea nodului rdcin!
-1

La acest nivel se calculeaz un maxim. Acest maxim (valoarea nodului rdcin) nu

poate fi mai mic dect -1!

La acest nivel se calculeaz un minim. Orice valoare a nodului printe poate fi

mai mic sau egal cu -1.

Metoda alpha-beta
1 -1 -2 -1

MA X MIN
2

0 -1 1 0

-1

0 -2 -1 0

Metoda alpha-beta
function alpha-beta(state, player, depth) begin if (depth = 0) then return score(state); val = worst(player); while (mai sunt stri de generat) begin generez o stare -> s; newval <- alpha-beta(s, not(player), depth-1); if player=MAX & newval val then return(newval); else if player=MIN & newval val then return(newval); else val back-up-compare(val, min-max(s, not(player), depth-1), player); // urmtoarea micare micoreaz spaiul de cutare n cazul n care se obine poziia de ctig // ntr-una din strile generate: if (val = -worst(player)) return(val); end return(val); end function worst(player) begin if player = MAX then return - ; else return + ; end function back-up-compare(val1, val2, player) begin if player = MAX then return max(val1, val2); else return min(val1, val2); end

Apelul:

alpha-beta (

,MAX,2)

Curs 11
Reprezentarea cunoaterii prin reele semantice evenimeniale

Reele semantice descriptive

adecvate reprezentrii cunoaterii statice se descriu:
entiti, n ierarhia de la general spre specific relaii ntre entiti

dou niveluri:
conceptual (intensiv): concepte (tipuri) referenial (extensiv): instane ale conceptelor

Reele semantice evenimeniale

adecvate reprezentrii cunoaterii dinamice se descriu:
entiti i tipuri evenimente n care sunt angrenate entitile reguli de bun-sim secvene de evenimente

pot fi folosite la:

explicarea semnificaiei unor enunuri generarea de situaii specifice pentru verificarea unor condiii simularea comportamentului unor societi de ageni

Reprezentri evenimeniale
Ion a spat asear gradina cu hrleul.
persoan a spa grdin

hrle Ion

ISA

ISA hrle1

ISA AG

ISA OB grdina1

INSTR

ev1

TMP

asear

Reguli de modelare a lumii reale

persoan REC X obiect REC Y AG cause a scpa REC a cdea AG MOD repede X-obiect REC plausible a cdea a atinge Y-obiect

repede

dur ATR Y-obiect

MOD
a atinge AG ISA AG X-obiect a lovi REC Y-obiect X-obiect AG fragil ATR

UPON REC plausible a lovi REC a se sparge

REC

Y-lichid Z-object CONTAINS UPON REC X-container a se sparge cause ON

REC

Z-obiect ON
ON a se revrsa Y-lichid REC cause a se revrsa REC FROM

a fi ntins

 Maria a scpat oul din mn. Ea a curat apoi pardoseala

t0
ev1 C

timp

1. Maria a scpat oul din mn.

persoan a scpa ISA ISA Maria PART-OF mn-1 ISA mn fragil lichid-1 ISA lichid REC FROM ou-1 ATR CONTAINS AG ev1

Reguli de modelare a lumii reale

persoan REC X obiect REC Y AG cause a scpa REC a cdea AG MOD repede X-obiect REC plausible a cdea a atinge Y-obiect

repede

dur ATR Y-obiect

MOD
a atinge AG ISA AG X-obiect a lovi REC Y-obiect X-obiect AG fragil ATR

UPON REC plausible a lovi REC a se sparge

REC

Y-lichid Z-object CONTAINS UPON REC X-container a se sparge cause ON

REC

Z-obiect ON
ON a se revrsa Y-lichid REC cause a se revrsa REC FROM

a fi ntins

t0
ev1 C ev2

timp

1. Maria a scpat oul din mn.

persoan a cdea ISA ISA Maria ev2 PART-OF mn-1 ou-1 ISA mn fragil lichid-1 ISA lichid ATR CONTAINS REC

Reguli de modelare a lumii reale

persoan REC X obiect REC Y AG cause a scpa REC a cdea AG MOD repede X-obiect REC plausible a cdea a atinge Y-obiect

repede

dur ATR Y-obiect

MOD
a atinge AG ISA AG X-obiect a lovi REC Y-obiect X-obiect AG fragil ATR

UPON REC plausible a lovi REC a se sparge

REC

Y-lichid Z-object CONTAINS UPON REC X-container a se sparge cause ON

REC

Z-obiect ON
ON a se revrsa Y-lichid REC cause a se revrsa REC FROM

a fi ntins

t0
ev1 C ev2 ev3

t3

timp

1. Maria a scpat oul din mn.

persoan a atinge ISA ISA Maria ev3 PART-OF REC mn-1 ou-1 ISA mn fragil lichid-1 ISA ?? ATR CONTAINS AG REC X-obiect ISA MOD repede

lichid

Reguli de modelare a lumii reale

persoan REC X obiect REC Y AG cause a scpa REC a cdea AG MOD repede X-obiect REC plausible a cdea a atinge Y-obiect

repede

dur ATR Y-obiect

MOD
a atinge AG ISA AG X-obiect a lovi REC Y-obiect X-obiect AG fragil ATR

UPON REC plausible a lovi REC a se sparge

REC

Y-lichid Z-object CONTAINS UPON REC X-container a se sparge cause ON

REC

Z-obiect ON
ON a se revrsa Y-lichid REC cause a se revrsa REC FROM

a fi ntins

t0
ev1 C ev2 ev3

t3

timp

1. Maria a scpat oul din mn.

persoan a lovi ISA ISA Maria ev3 PART-OF mn-1 ou-1 ISA mn fragil lichid-1 ISA lichid ?? ATR CONTAINS AG REC X-obiect ISA

Reguli de modelare a lumii reale

persoan REC X obiect REC Y AG cause a scpa REC a cdea AG MOD repede X-obiect REC plausible a cdea a atinge Y-obiect

repede

dur ATR Y-obiect

MOD
a atinge AG ISA AG X-obiect a lovi REC Y-obiect X-obiect AG fragil ATR

UPON REC plausible a lovi REC a se sparge

REC

Y-lichid Z-object CONTAINS UPON REC X-container a se sparge cause ON

REC

Z-obiect ON
ON a se revrsa Y-lichid REC cause a se revrsa REC FROM

a fi ntins

t0
ev1 C ev2 ev3

t3
ev4

timp

1. Maria a scpat oul din mn.

persoan a lovi ISA ISA Maria ev3 PART-OF mn-1 ou-1 ISA mn fragil lichid-1 ISA lichid ?? ATR CONTAINS AG REC X-obiect ISA UPON REC ISA se sparge

ev4

Reguli de modelare a lumii reale

persoan REC X obiect REC Y AG cause a scpa REC a cdea AG MOD repede X-obiect REC plausible a cdea a atinge Y-obiect

repede

dur ATR Y-obiect

MOD
a atinge AG ISA AG X-obiect a lovi REC Y-obiect X-obiect AG fragil ATR

UPON REC plausible a lovi REC a se sparge

REC

Y-lichid Z-object CONTAINS UPON REC X-container a se sparge cause ON

REC

Z-obiect ON
ON a se revrsa Y-lichid REC cause a se revrsa REC FROM

a fi ntins

t0
ev1 C ev2 ev3

t3
ev4 ev5

timp

1. Maria a scpat oul din mn.

persoan se revars ISA ISA Maria ev5 PART-OF mn-1 ou-1 ISA mn fragil lichid-1 ISA lichid ?? ATR CONTAINS X-obiect ISA FROM REC RECO N

Reguli de modelare a lumii reale

persoan REC X obiect REC Y AG cause a scpa REC a cdea AG MOD repede X-obiect REC plausible a cdea a atinge Y-obiect

repede

dur ATR Y-obiect

MOD
a atinge AG ISA AG X-obiect a lovi REC Y-obiect X-obiect AG fragil ATR

UPON REC plausible a lovi REC a se sparge

REC

Y-lichid Z-object CONTAINS UPON REC X-container a se sparge cause ON

REC

Z-obiect ON
ON a se revrsa Y-lichid REC cause a se revrsa REC FROM

a fi ntins

t0
ev1 C ev2 ev3

t3
ev4

t6
ev5 ev6

timp

1. Maria a scpat oul din mn.

persoan se revars ISA ISA Maria ev5 PART-OF mn-1 ou-1 ISA mn fragil lichid-1 ISA lichid ?? ATR CONTAINS X-obiect ISA FROM REC REC REC ON ISA ev6 a fi ntins

Reguli de modelare a proceselor cognitive

persoan lichid persoan ISA X-persoan X-persoan X-persoan AG plausible crede cura crede

ISA
X-persoan

ISA Y-lichid

AG OB vedea

REC

AG

AG

plausible
a fi ntins

ON HAS-PROP Z-obiect Z-obiect murdar

OB HAS-PROP Y-obiect Y-obiect

murdar

Reguli de modelare a proceselor cognitive

persoan lichid persoan ISA X-persoan X-persoan X-persoan AG plausible crede cura crede

ISA
X-persoan

ISA Y-lichid

AG OB vedea

REC

AG

AG

plausible
a fi ntins

ON HAS-PROP Z-obiect Z-obiect murdar

OB HAS-PROP Y-obiect Z-obiect

murdar

t0
ev1 C ev2 ev3

t3
ev4

t6
ev5 ev6 ev7

timp

1. Maria a scpat oul din mn.

persoan a fi ntins ISA ISA Maria PART-OF OB mn-1 ou-1 ISA mn fragil lichid-1 ISA lichid ?? ATR CONTAINS X-obiect ISA REC ON X-obiect ev6 AG ev7 ISA

crede

HAS-PROP
murdar

t0
ev1 C ev2 ev3

t3
ev4

t6
ev5 ev6 ev7

t8
ev8

timp

1. Maria a scpat oul din mn.

persoan a fi ntins ISA ISA Maria PART-OF OB mn-1 ou-1 ISA mn fragil lichid-1 ISA lichid ?? ATR CONTAINS X-obiect ISA REC ON X-obiect ev6 AG ev7 ISA

crede

HAS-PROP
murdar

2. Ea a curat apoi pardoseala.

cura

ISA AG ev8

OB

pardoseala

Reguli de modelare a proceselor cognitive

persoan lichid persoan ISA X-persoan X-persoan X-persoan AG plausible crede cura crede

ISA
X-persoan

ISA Y-lichid

AG OB vedea

REC

AG

AG

plausible
a fi ntins

ON HAS-PROP Z-obiect Z-obiect murdar

OB HAS-PROP Y-obiect Z-obiect

murdar

t0
ev1 C ev2 ev3

t3
ev4

t6
ev5 ev6 ev7

t8
ev9 ev8

timp

1. Maria a scpat oul din mn.

persoan

matching

a fi ntins

crede

ISA ISA Maria PART-OF OB mn-1 ou-1 ISA mn fragil lichid-1 ISA lichid ?? ATR CONTAINS X-obiect ISA REC ON X-obiect ev6 AG ev7 ISA

HAS-PROP
murdar

Se potrivesc!

2. Ea a curat apoi pardoseala.

cura crede

ISA AG ev8 AG ev9

ISA

OB

OB HAS-PROP

pardoseala

pardoseala

murdar

t0
ev1 C ev2 ev3

t3
ev4

t6
ev5 ev6 ev7

t8
ev9 ev8

timp

1. Maria a scpat oul din mn.

persoan a scpa ISA ISA Maria PART-OF mn-1 ISA mn fragil lichid-1 pardoseala ISA lichid murdar REC FROM ou-1 ATR CONTAINS X-obiect ISA FROM REC REC REC AG REC AG ev1 ISA REC ev2 ev3 UPON ON OB ON X-obiect ISA ISA ISA ev5 ISA ev6 AG ISA ev7 a cdea a lovi se sparge se revars a fi ntins

crede

ev4

HAS-PROP

2. Ea a curat apoi pardoseala.

cura crede

nelegem...

ISA AG ev8 AG ev9

ISA

OB

OB HAS-PROP

pardoseala

pardoseala

murdar