Knowledge Repr e sent at i on
-
Upload
gretchen-mcintosh -
Category
Documents
-
view
24 -
download
1
description
Transcript of Knowledge Repr e sent at i on
Knowledge Representation
1
Outline
2
General ontology Categories and objects Events and processes Reasoning systems Internet shopping world Summary
Ontologies
3
An ontology is a “vocabulary” and a “theory” of acertain “part of reality”
Special-purpose ontologies apply to restricteddomains (e.g. electronic circuits)
General-purpose ontologies have wider applicability across domains, i.e.
Must include concepts that cover many subdomains Cannot use special “short-cuts” (such as ignoring time) Must allow unification of different types of knowledge
GP ontologies are useful in widening applicabilityof reasoning systems, e.g. by including time
Ontological engineering
4
Representing a general-purpose ontology is adifficult task called ontology engineering
Existing GP ontologies have been created indifferent ways:
By team of trained ontologists By importing concepts from database(s) By extracting information from text documents By inviting anybody to enter commonsense knowledge
Ontological engineering has only been partially successful, and few large AI systems are based on GP ontologies (use special purpose ontologies)
Elements of a general ontology
5
Categories of objects Measures of quantities Composite objects Time, space, and change Events and processes Physical objects Substances Mental objects and beliefs
Top-level ontology of the world
6
Anything
EventsAbstractObjects
PhysicalObjectsPlacesIntervalsRepresentationObjectsSetsNumbers
Processes
Categories Sentences Measurements Moments
StuffThings
Liquid GasAgents SolidsAnimalsWeights
Times
Humans
Upper Ontology
• The general framework of concepts is called an upper ontology because of the convention of drawing graphs with the general concepts at the top and the more specific concepts below them• Of what use is an upper ontology?
• Consider the ontology for circuits that we studied• It makes many simplifying assumptions: time is omitted completely; signals
are fixed and do not propagate; the structure of the circuit remains constant.• more general ontology would consider signals at particular times, and would
include the wire lengths and propagation delays.• This would allow us to simulate the timing properties of the circuit, and
indeed such simulations are often carried out by circuit designers.
7
Categories and objects
8
Categories are used to classify objectsaccording to common properties or definitions
x x Tomates Re d (x ) Round (x )
Categories can be represented by Predicates: Tomato(x) Objects: The constant Tomatoes represents set of
tomatoes (reification) Roles of category representations Instance relations (is-a):
Taxonomical hierarchies (Subset): Tom atoes Fruit
Inheritance of properties (Exhaustive) decompositions
x1 Tomatoes
Categories using FOL
9
Properties of categories
• We say that two or more categories are disjoint if they have no members in common.• exhaustive decomposition• A disjoint exhaustive decomposition is known as a partition. • The following examples illustrate these three concepts:
• The predicates us to define these concepts are
• For example, a bachelor is an unmarried adult male:10
Objects and substance
11
Need to distinguish between substance anddiscrete objects
Substance (“stuff”) Mass nouns - not countable Intrinsic properties Part of a substance is (still) the same substance
Discrete objects (“things”) Count nouns - countable Extrinsic properties Parts are (generally) not of same category
Composite objects
12
A composite object is an object that has otherobjects as parts
The PartOf relation defines the object containment, and is transitive and reflexive PartOf (x, y) PartOf ( y, z) PartOf (x, z)
PartOf (x, x)
Objects can be grouped in PartOf hierarchies,similar to Subset hierarchies
The structure of the composite objectdescribes how the parts are related
Composite objects• For example, a biped has two legs attached to a body:
• For example, we might want to say “The apples in this bag weigh two pounds.”• we need a new concept, which we will call a bunch.• For example, if the apples are Apple1, Apple2, and Apple3, then
BunchOf ({Apple1,Apple2,Apple3})∀x x s PartOf (x, BunchOf (s))∈ ⇒
∀ y [ x x s PartOf (x, y)] PartOf (BunchOf (s), y) ∀ ∈ ⇒ ⇒
• logical minimization, which means defining an object as the smallest one satisfying certain conditions.
13
Measurements
14
Need to be able to represent properties like height, mass, cost, etc. Values for such properties are measures
Unit functions represent and convert measures
Length( L1) Inches(1.5) Centimeters(3.81)
l Centimeters(2.54 l) Inches(l) Measures can be used to describe objects
Mass(Tomato1) Ki log rams(0.16)
d d Days Duration(d ) Hours(24) Non-numerical measures can also be represen-
ted, but normally there is an order (e.g. >). Used in qualitative physics
Measurements
• Comparative difficulty
e1 Exercises e2 Exercises Wrote(Norvig, e1) Wrote(Russell, ∈ ∧ ∈ ∧ ∧e2) Difficulty(e1) > Difficulty(e2) ⇒
e1 Exercises e2 Exercises Difficulty(e1) > Difficulty(e2) ∈ ∧ ∈ ∧ ⇒ExpectedScore(e1) < ExpectedScore(e2)
15
Objects: Things and stuff
• The real world can be seen as consisting of primitive objects (e.g., atomic particles) and composite objects built from them.• There is, however, a significant portion of reality that seems to defy
any obvious individuation—division into distinct objects. We give this portion the generic name stuff.• count nouns, such as aardvarks, holes, and theorems, and mass
nouns, such as butter, water, and energy.• To represent stuff properly, we begin with the obvious. We need to
have as objects in our ontology at least the gross “lumps” of stuff we interact with.
b Butter PartOf (p, b) p Butter ∈ ∧ ⇒ ∈b Butter MeltingPoint(b,Centigrade(30))∈ ⇒
• Intrinsic properties and extrinsic properties
16
Event calculus
17
Event calculus: How to deal with change based onrepresenting points of time
Reifies fluents and events A fluent: At(Bilal, Berkeley) The fluent is true at time t: T(At(Bilal, IQRA),t)
Events are instances of event categoriesE1 Flyings Flyer(E1, Bilal) Origin(E1, SF ) Destination(E1, KHI)
Event E1 took place over interval i Happens(E1, i)
Time intervals represented by (start, end) pairs i = (t1, t2)
Event calculus predicates
18
T(f, t) Happens(e, i) Initiates(e, f, t)
Fluent f is true at time tEvent e happens over interval iEvent e causes fluent f to start at t
Terminates(e, f, t) Event e causes f to cease at t Clipped(f, t) Restored(f, i)
Fluent f ceases to be true in int. iFluent f becomes true in interval i
Events
• We assume a distinguished event, Start , that describes the initial state by saying which fluents are initiated or terminated at the start time.• We define T by saying that a fluent holds at a point in time if the fluent
was initiated by an event at some time in the past and was not made false (clipped) by an intervening event.• A fluent does not hold if it was terminated by an event and not made
true (restored) by another event.• Happens(e, (t1, t2)) Initiates(e, f, t1) ∧ ∧ ¬ Clipped(f, (t1, t)) t1 < t T(f, t)∧ ⇒• Happens(e, (t1, t2)) Terminates(e, f, t1) ∧ ∧ ¬ Restored (f, (t1, t)) t1 < t ∧ ⇒ ¬
T(f, t)
where Clipped and Restored are defined by• Clipped(f, (t1, t2)) ⇔ e, t, t3 Happens(e, (t, t3)) t1 ≤ t < t2 Terminates(e, f, ∃ ∧ ∧
t)• Restored (f, (t1, t2)) ⇔ e, t, t3 Happens(e, (t, t3)) t1 ≤ t < t2 Initiates(e, f, t)∃ ∧ ∧
19
Processes
• The events we have seen so far are what we call discrete events• Categories of events with sub-intervals are called process categories
or liquid event categories
20
Time intervals
21
Time intervals are partitioned into moments (zero duration) and extended intervals
Partition(Moments,ExtendedIntervals,Intervals)
i i Intervals (i Moments Duration(i) 0) Functions Start and End delimit intervals
i Interval(i) Duration(i) (Time(End (i)) Time(Start (i))) May use e.g. January 1, 1900 as arbitrary time 0
Time(Start(AD1900))=Seconds(0)
Relations between time intervals
22
Meet(i, j)
Before(i, j)After(j, i)
During(i, j )
Overlap(i, j)Overlap(j, i)
j
i
j
ij
i
j
Can be expressed logically, e.g.
i , j Meet (i, j) Time(End (i)) Time(Start ( j))
Mental events and mental objects
Need to represent beliefs in self and other agents, e.g. for controlling reasoning, or for planning actions that involve others
How are beliefs represented?• Beliefs are reified as mental objects• Mental objects are represented as strings in a
language• Inference rules for this language can be defined
Rules for reasoning about logical agents’ use their beliefs• a , p ,q LogicalAgent (a) Believes(a , p)
• Believes(a ," p q") Believes(a ,q)
23
a , p LogicalAgent (a) Believes(a , p)
Believes(a ,"Believes( Name(a), p)" )
Mental events
• propositional attitudes that an agent can have toward mental objects: attitudes such as Believes, Knows, Wants, Intends, and Informs• For example, suppose we try to assert that Lois knows that Superman
can fly:Knows(Lois, CanFly(Superman))
• if it is true that Superman is Clark Kent, then we must conclude that Lois knows that Clark can fly:
(Superman = Clark) Knows(Lois , CanFly(Superman)) |= Knows(Lois, ∧CanFly(Clark ))
• This property is called referential transparency
24
Modal Logic
• Modal logic is designed to address this problem. • Regular logic is concerned with a single modality, the modality of
truth, allowing us to express “P is true.” • Modal logic includes special modal operators that take sentences
(rather than terms) as arguments.• For example, “A knows P” is represented with the notation KAP,
where K is the modal operator for knowledge. It takes two arguments, an agent (written as the subscript) and a sentence.
25
Semantic networks
26
Graph representation of categories, objects,relations, etc. (i.e. essentially FOL)
Natural representation of inheritance and default values
∀x x∈ Persons ⇒ [∀ y HasMother (x, y) ⇒ y ∈ FemalePersons ] .
∀x x∈ Persons ⇒ Legs(x, 2) .
Semantic Network
Joe
Boy
Kay
Woman
Food
HumanBeing
School
Hasa child
NeedsGoes to
Is a
Is a
Is a
Is a
27
Other reasoning systems for categories
28
Description logics Derived from semantic networks, but more formal Supports subsumption, classification and consistency
Circumscription and default logic Formalizes reasoning about default values Assumes default in absence of other input; must be
able to retract assumption if new evidence occurs Truth maintenance systems
Supports belief revision in systems where retracting belief is permitted
Internet shopping world
29
An agent that understands and acts in aninternet shopping environment
The task is to shop for a product on the Web,given the user’s product description
The product description may be precise, in which case the agent should find the best price
In other cases the description is only partial, and the agent has to compare products
The shopping agent depends on having productknowledge, incl. category hierarchies
PEAS specification of shopping agent
30
Performance goal Recommend product(s) to match user’s description
Environment All of the Web
Actions Following links Retrieve page contents
Sensors Web pages: HTML, XML
Outline of agent behavior
31
Start at home page of known web store(s) Must have knowledge of relevant web addresses,
such as www.amazon.com etc. Spread out from home page, following links to
relevant pages containing product offers Must be able to identify page relevance, using
product category ontologies, as well parse page contents to detect product offers
Having located one or more product offers,agent must compare and recommend product
Comparison range from simple price ranking to
complex tradeoffs in several dimensions
Following links
• The agent will have knowledge of a number of stores, for example:Amazon OnlineStores Homepage(Amazon, “amazon.com”) .∈ ∧
Ebay OnlineStores Homepage(Ebay, “ebay.com”) .∈ ∧ExampleStore OnlineStores Homepage(ExampleStore, “example.com”)∈ ∧• a page is relevant to the query if it can be reached by a chain of zero or
more relevant category links from a store’s home page, and then from one more link to the product offer.
Relevant(page, query) store, home store OnlineStores ⇔ ∃ ∈ ∧Homepage(store, home) url , url 2 RelevantChain(home, url 2, query) ∧ ∃ ∧
Link(url 2, url ) page = Contents(url ) ∧
RelevantChain(start , end, query) (start = end) ( u, text LinkText(start, ⇔ ∨ ∃u, text ) RelevantCategoryName(query, text ) RelevantChain(u, end, ∧ ∧
query)) . 32
Following Links
33
Comparing offers
∃ c, offer c LaptopComputers offer ProductOffers ∈ ∧ ∈ ∧Manufacturer(c,IBM ) Model (c, ThinkBook970 ) ∧ ∧ScreenSize(c, Inches(14)) ScreenType(c, ColorLCD) ∧ ∧MemorySize(c,Gigabytes(2)) CPUSpeed (c,GHz (1.2)) ∧ ∧OfferedProduct(offer, c) Store(offer , GenStore) ∧ ∧URL(offer , “example.com/computers/34356.html”) ∧Price(offer , $(399)) Date(offer ,Today)∧
34
Summary
35
An ontology is an encoding of vocabulary and relationships. Special-purpose ontologies can be effective within limited domains
A general-purpose ontology needs to cover a wide variety of knowledge, and is based on categories and an event calculus
It covers structured objects, time and space, change, processes, substances, and beliefs
The general ontology can support agent reasoning in a wide variety of domains, including the Internet shopping world