University Politehnica of Bucharest2008-2009

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

Artificial IntelligenceArtificial Intelligence

2

Structured knowledge representationStructured knowledge representation

Semantic NetworksSemantic Networks UnitsUnits Specific inferencesSpecific inferences Problems with inheritanceProblems with inheritance Semantic WebSemantic Web

Lecture No. 8

3

Semantic networks

first model for structured knowledge representation

describe semantics of natural language used extensively as a model for

representing knowledge in KBS

Knowledge base Radu sent a letter to Maria. Radu is a student. Ioana is a schoolgirl. Radu's address is Luterana, 15.

occupation (radu, student) occupation (ioana, schoolgirl) send (radu, ioana, letter) address (radu, luterana - 15)

Group knowledge into entities Radu

– occupation (radu, student)– send (radu, ioana, letter)– address (radu, luterana - 15)

Ioana– occupation (ioana, schoolgirl)– send (radu, ioana, letter)

Radu occupation: student

address: luterana-15

Ioanaoccupation: schoolgirl

Associate propertiesor attributes

Model 2 args predicates

What about 3 args predicates?

radu),Sender(t

event)-sending,ISA(t

1

1

letter),Object(t

ioana),Receiver(t

1

1

radu)(x,Sevent)-sendingx)(ISA(x,( ender

letter))Object(x,ioana),Receiver(x

- use ISA predicate- use Skolemizationto eliminateexistential quantifier

7

RaduISA: PersonOccupation: studentAddress: luterana-15

IoanaISA: Person Occupation : schoolgirl

T1 ISA: Sending-event Sender: Radu

Receiver: Ioana Object: letterslots

8

ISA introduces a class-instance relationshipAKO predicate introduces a subclass-class relationship

(x)) (x)event -(Sending x)( Event

(x)) (Person(x) x)( gLivingThin

9

Sending-eventAKO: EventSender: PersonReceiver: PersonObject: ObjectClass

PersonAKO: living ThingOccupation: (student, engineer,

…)Address: string

10

Canar galben

aripi

Pasare

culoare

ISA

are

zbura

poate

(a)

Eveniment

Eveniment-trimitere

AKO

T1

ISA

scrisoare

Obiect

Radu

Expeditor

Ioana

Destinatar

Persoana

ISA

ISA

Adresaluterana-15

student

Ocupatie

elev

(b)

Ocupatie

Examples of SNExamples of SN

11

Units MemberOf - ISA SubClass, SuperClass - AKO

UnitsUnits

12

Facets– Features associated to slots– Possible facets

Value facet Value type facet Default facet If-needed procedure or function Demon facet Comment facet

FacetsFacets

Specific inferences in semantic networks: properties/attributes inheritance

1) Inheritance of attributes along the ISA relation 2) Inheritance of attributes along the AKO

relation

Attribute's value inheritance

Piramida Caramida

Bloc

AKO AKO

triunghi dreptunghiForma Forma

Consistentamare

Piramida18 Caramida12

ISA ISA

M o s t e n i r e a v a l o r i l o r i n r e t e l e s e m a n t i c eAttribute's value inheritance

Algorithm: Inheritance of attribute's value in a class hierarchyThe algorithm gets the value V of an attribute A of the object O

FindVal (O, A, V)1. Create a list L with node O and all the nodes linked to O by an ISA relationship2. while L != [ ] do

2.1. Remove first node N from L 2.2. if attribute A of node N has value, be it V

then 2.2.1. Place V as the value of attribute A of object O2.2.3. return SUCCESS

2.3. Add all nodes linked by AKO to N at the end of L3. return FAILend.

16

Caramida rosieCuloare

Caramida12

ISAFateta valoare implicita

Caramida0 albaCuloare

ISA

Mostenirea valorilor implicite in retele semanticeAttribute's default value inheritance

Algorithm: Inheritance of attribute's default value in a class hierarchy

The algorithm gets the default value V of an attribute A of the object OFindDefault (O, A, V)1. Create a list L with node O and all the nodes linked to O by an ISA relationship2. while L != [ ] do

2.1. Remove first node N from L 2.2. if attribute A of node N has a default value, be it V

then 2.2.1. Place V as the value of attribute A of object O2.2.3. return SUCCESS

2.3. Add all nodes linked by AKO to N at the end of L3. return FAILend.

FindIfNeeded(O,A,V)

18

Control strategyControl strategy the control strategy of inheritance indicates

the order of considering the different facets

2 basic strategies– N strategy– Z strategy

Algorithm: The algorithm gets the value V of an attribute A of the object O using the N strategy

FindValN (O, A, V)1. if FindVal (O,A,V) = SUCCESS

then return SUCCESS2. if FindDefault (O,A,V) = SUCCESS

then return SUCCESS3. if FindIfNeeded (O,A,V) = SUCCESS

then return SUCCESS4. return FAILend

Strategy NStrategy N

Algoritm: Strategia Z de determinare a valorii unui atribut.Algoritmul determina valoarea unui atribut A al unei instante O utilizind strategia Z.

DetValZ (O, A, V)1. Formeaza o lista L cu nodul O si toate nodurile

legate de O prin relatia ISA2. cat timp L != [ ] executa

2.1. Elimina primul nod, N, din lista L2.2. daca fateta valoare a atributului A a

nodului N este V

then 2.2.1. Depune V in nodul punctat de

atributul A al obiectului O2.2.2. intoarce SUCCES

Strategy ZStrategy Z

2.3. daca fateta valoare implicita a atributului A a nodului N este Vatunci 2.3.1. Depune V in nodul punctat de atributul A al

obiectului O2.3.2. intoarce SUCCES

2.4. daca fateta procedura necesara a atributului A a nodului N este proc (A1,..., An,V)atunci2.4.1. Determina valorile atributelor A1,..., An ale

instantei O2.4.2. daca s-au gasit valori pentru A1,..., An

atuncii. executa proc (A1,..., An,V)ii. Depune V in nodul punctat de

atributul A al obiectului Oiii. intoarce SUCCES

3. intoarce INSUCCESsfarsit.

22

Problems with inheritance

DAG networks

Multiple inheritance

Unit PasareSlot: Zboara

Value: da

Unit FifiSlot: Zboara

Value: necunoscut

Unit StrutSlot: Zboara

Value: nu

Unit PasareZoo

ISA (MemberOf) ISA (MemberOf)

AKO (SuperClasses)

Zboara Fifi?

AKO (SuperClasses)

DAG Semantic network

24

Multiple inheritanceMultiple inheritance

Distance between units If we look for the value of an attribute A of a

unit U1 and find 2 paths : U1U2 and U1U3 to U2 and U3 containing values for A

then compute the length of the alternate paths from the instance/unit U1 to units U2 and U3 and choose the value from the closest unit (shortest path)

May cause semantic problems

25

Unit PasareSlot: Zboara

Value: da

Unit FifiSlot: Zboara

Value: necunoscut

Unit StrutSlot: Zboara

Value: nu

Unit PasareZoo

ISA

ISA

AKO

Zboara Fifi? nu

Unit StrutPenat

Unit StrutPenatAlb

AKO

AKO

AKO

Cea mai micadistanta

Cea mai micadistanta inferentiala

D istanta si d istanta in ferentia la in tre un ita tiDistance and inferential distance

Shortestdistance

Shortestinferentialdistance

26

Multiple inheritanceMultiple inheritance

Inferential distance

Unit1 is closer to Unit2 than to Unit3 if and only if Unit1 has an inferential path to Unit3 which contains Unit2.

Unit1 is closer to Unit2 than to Unit3 if and only if Unit2 is on an inferential path of ISA and AKO relationships between Unit1 and Unit3

Algorithm: Inheritance of attribute's value in a class hierarchy based on inferential distanceFind the value V of slot S of unit U

1. Make a list L with unit U and all units to which unit U is linked by an ISA/MemberOf

2. Initialize a list CAND = [ ] 3. while L != [ ] do

3.1. remove first unit X from L3.2. if slot S of X has value

then CAND = CAND {X}3.3. else add to L all units to which X is linked by a

AKO/SuperClass4. for each unit C CAND do

4.1. Verify if there is another C’ CAND having an inferential distance to U shorter than to C

4.2. if such a C' existsthen remove C from CAND

5. if card (CAND) = 0then return FAIL /* no value for S */

6. if card (CAND) = 1then 6.1. Be C the one element of CAND6.2. Make the value V of slot S of C the value of slot

S of U6.3. return SUCCESS

7. if card (CAND) > 1 then return CONTRADICTION

end.

Combined representations

Knowledge base contains:- declarative knowledge: units

- procedural knowledge: rules daca Camion.Inaltime > 2 si Camion.Culoare = rosuatunci Camion.Pret = 1000

Inferences: both specific to units and specific to rules

30

Semantic Web and Ontologies

Semantic Web Web was “invented” by Tim Berners-Lee (amongst others),

a physicist working at CERN TBL’s original vision of the Web was much more ambitious

than the reality of the existing (syntactic) Web:

TBL (and others) have since been working towards realising this vision, which has become known as the Semantic Web– E.g., article in May 2001 issue of Scientific American…

“... a goal of the Web was that, if the interaction between person and hypertext could be so intuitive that the machine-readable information space gave an accurate representation of the state of people's thoughts, interactions, and work patterns, then machine analysis could become a very powerful management tool, seeing patterns in our work and facilitating our working together through the typical problems which beset the management of large organizations.”

Where we are Today: the Syntactic Web

[Hendler & Miller 02]

The Syntactic Web is… A hypermedia, a digital library

– A library of documents called (web pages) interconnected by a hypermedia of links

A database, an application platform– A common portal to applications accessible through web pages,

and presenting their results as web pages A platform for multimedia

– BBC Radio 4 anywhere in the world! Terminator 3 trailers! A naming scheme

– Unique identity for those documents

A place where computers do the presentation (easy) and people do the linking and interpreting (hard).

Why not get computers to do more of the hard work?

[Goble 03]

Impossible (?) using the Syntactic Web… Complex queries involving background knowledge

– Find information about “animals that use sonar but are not either bats or dolphins”

Locating information in data repositories– Travel enquiries– Prices of goods and services– Results of human genome experiments

Delegating complex tasks to web “agents”– Book me a holiday next weekend somewhere warm, not too

far away, and where they speak French or English

What is the Problem? Consider a typical web page:

Markup consists of: – rendering

information (e.g., font size and colour)

– Hyper-links to related content

Semantic content is accessible to humans but not (easily) to computers…

What information can we see…WWW2002The eleventh international world wide web conferenceSheraton waikiki hotelHonolulu, hawaii, USA7-11 may 20021 location 5 days learn interactRegistered participants coming fromaustralia, canada, chile denmark, france, germany, ghana, hong kong, india,

ireland, italy, japan, malta, new zealand, the netherlands, norway, singapore, switzerland, the united kingdom, the united states, vietnam, zaire

Register nowOn the 7th May Honolulu will provide the backdrop of the eleventh

international world wide web conference. This prestigious event …Speakers confirmedTim berners-lee Tim is the well known inventor of the Web, …Ian FosterIan is the pioneer of the Grid, the next generation internet …

What information can a machine see…

…

…

…

Solution: XML markup with “meaningful” tags?<name> </

name><location> </location>

<date> </date><slogan> </slogan><participants>

</participants>

<introduction>

…

</introduction><speaker> </speaker><bio> </bio>…

But What About…<conf> </

conf><place> </place>

<date> </date><slogan> </slogan><participants>

</participants>

<introduction>

…

</introduction><speaker> </speaker><bio> …

Must add semantics External agreement on meaning of annotations

– E.g., Dublin Core• Agree on the meaning of a set of annotation tags

– Problems with this approach• Inflexible• Limited number of things can be expressed

Use Ontologies to specify meaning of annotations– Ontologies provide a vocabulary of terms– New terms can be formed by combining existing ones– Meaning (semantics) of such terms is formally specified– Can also specify relationships between terms in multiple

ontologies

Ontology in philosophy

a philosophical discipline—a branch of philosophy that deals with the nature and the organisation of reality

Science of Being (Aristotle, Metaphysics, IV, 1)

Tries to answer the questions:

What characterizes being?

Eventually, what is being?

Ontology in Computer Science

An ontology is an engineering artifact: – It is constituted by a specific vocabulary used to describe a

certain reality, plus – a set of explicit assumptions regarding the intended meaning

of the vocabulary.

Thus, an ontology describes a formal specification of a certain domain:– Shared understanding of a domain of interest– Formal and machine manipulable model of a domain of

interest

“An explicit specification of a conceptualisation” [Gruber93]

Structure of an ontology

Ontologies typically have two distinct components:

Names for important concepts in the domain– Elephant is a concept whose members are a kind of animal– Herbivore is a concept whose members are exactly those

animals who eat only plants or parts of plants – Adult_Elephant is a concept whose members are exactly

those elephants whose age is greater than 20 years

Background knowledge/constraints on the domain– Adult_Elephants weigh at least 2,000 kg– All Elephants are either African_Elephants or

Indian_Elephants– No individual can be both a Herbivore and a Carnivore

Example of ontology

Tools

Description Logics OWL = Web Ontological Language Protégé – framework for building

ontologies