Knowledge Representation

What to Represent?
Let us first consider what kinds of knowledge
might need to be represented in AI systems:

Objects
 -- Facts about objects in our world
domain. e.g. Guitars have strings, trumpets are brass
instruments.
Events
 -- Actions that occur in our world. e.g. Steve Vai
played the guitar in Frank Zappa's Band.
Performance
 -- A behavior like playing the guitar involves
knowledge about how to do things.

Meta-knowledge
 -- knowledge about what we know.
Thus in solving problems in AI we must
represent knowledge and there are two
entities to deal with:
Facts
 -- truths about the real world and what we
represent. This can be regarded as
the knowledge level
Representation of the facts
 which we manipulate. This can be regarded as
the symbol level since we usually define the
representation in terms of symbols that can be
manipulated by programs.
We can structure these entities at two levels
the knowledge level
 -- at which facts are described
the symbol level
 -- at which representations of objects are
defined in terms of symbols that can be
manipulated in programs.
English or natural language is an obvious way
of representing and handling facts.
Logic enables us to consider the following
fact:
spot is a dog as dog(spot) We could then infer
that all dogs have tails with:
Vx : dog(x) hasatail(x) We can then deduce:
hasatail(Spot)
Using Knowledge (1)
We have briefly mentioned where knowledge is used
in AI systems. Let us consider a little further to what
applications and how knowledge may be used.

 Learning -- acquiring knowledge. This is more than

simply adding new facts to a knowledge base.
 New data may have to be classified prior to storage
for easy retrieval, etc.
 Duplication should be avoided.

 Reasoning -- Infer facts from existing data.

Using Knowledge(2)
If a system only knows:
Davis is a Jazz Musician.

All Jazz Musicians can play their instruments

well.

If things like Is Davis a Jazz Musician? or Can

Jazz Musicians play their instruments well? are
asked then the answer is readily obtained from
the data structures and procedures.

However a question like Can Davis play his

instrument well? requires reasoning.
Properties for Knowledge Representation Systems

The following properties should be possessed by a

knowledge representation system.
 Representational Adequacy
 -- the ability to represent the required knowledge;
 Inferential Adequacy
 - the ability to manipulate the knowledge represented to
produce new knowledge corresponding to that inferred
from the original;
 Inferential Efficiency
 - the ability to direct the inferential mechanisms into the
most productive directions by storing appropriate
guides;
 Acquisitional Efficiency
 - the ability to acquire new knowledge using automatic
methods wherever possible rather than reliance on
human intervention.

To date no single system optimises all of the above

Approaches to Knowledge
Representation
Simple relational knowledge

 The simplest way of storing facts is to use a relational

method where each fact about a set of objects is set out
systematically in columns. This representation gives
little opportunity for inference, but it can be used as the
knowledge basis for inference engines.

 Simple way to store facts.

 Each fact about a set of objects is set out systematically

in columns.

 Little opportunity for inference.

 Knowledge basis for inference engines.

Figure: Simple Relational
Knowledge
Inheritable knowledge

Relational knowledge is made up of objects

consisting of
attributes
corresponding associated values.
We extend the base more by allowing
inference mechanisms:
Property inheritance
 elements inherit values from being
members of a class.
 data must be organized into a hierarchy of
classes.
Boxed nodes -- objects and values of attributes of objects.
Values can be objects with attributes and so on.
Arrows -- point from object to its value.
This structure is known as a slot and filler structure,
semantic network or a collection of frames.
Inferential Knowledge
Represent knowledge as formal logic:
All dogs have tails
V(X) : dog(x)  hasatail(x)
Advantages:
A set of strict rules.
 Can be used to derive more facts.
 Truths of new statements can be verified.
 Guaranteed correctness.
Popular in AI systems. e.g Automated
theorem proving
Procedural Knowledge
Basic idea:
Knowledge encoded in some procedures
 small programs that know how to do specific
things, how to proceed.
Advantages:
Heuristic or domain specific knowledge can
be represented.
Extended logical inferences, such as default
reasoning facilitated.
Side effects of actions may be modeled. Some
rules may become false in time. Keeping
track of this in large systems may be tricky.
What is Learning?
Learning is an important area in AI, perhaps
more so than planning.
Problems are hard -- harder than planning.
A goal of AI is to enable computers that can
be taught rather than programmed.
Learning is a an area of AI that focuses on
processes of self-improvement.
Recognized Solutions are not as common as
planning.
Why is it hard?
Intelligence implies that an organism or
machine must be able to adapt to new
situations.
It must be able to learn to do new things.
This requires knowledge acquisition,
inference, updating/refinement of
knowledge base, acquisition of heuristics,
applying faster searches, etc.
How can we learn?

 Many approaches have been taken to

attempt to provide a machine with learning
capabilities. This is because learning tasks
cover a wide range of phenomena.

 Listed below are a few examples of how

one may learn.
Skill refinement -- one can learn by
practicing, e.g playing the piano.

Knowledge acquisition -- one can learn by

experience and by storing the experience in
a knowledge base. One basic example of this
type is rote learning.

Taking advice -- Similar to rote learning

although the knowledge that is input may
need to be transformed (or operationalised)
in order to be used effectively.
 Problem Solving -- if we solve a problem one
may learn from this experience. The next time we
see a similar problem we can solve it more
efficiently. This does not usually involve gathering
new knowledge but may involve reorganization of
data or remembering how to achieve to solution.
 Induction -- One can learn from examples.
 Discovery -- Here one learns knowledge without
the aid of a teacher.
 Analogy -- If a system can recognize similarities
in information already stored then it may be able
to transfer some knowledge to improve to
solution of the task in hand.
Rote Learning
Rote Learning is basically memorization.
 Saving knowledge so it can be used again.
 Retrieval is the only problem.
 No repeated computation, inference or query is
necessary.
A simple example of rote learning is caching
 Store computed values (or large piece of data)
 Recall this information when required by computation.
 Significant time savings can be achieved.
 Many AI programs (as well as more general ones)
have used caching very effectively.
Memorization is a key necessity for learning:
Store v Compute

Rote Learning must not decrease the efficiency

of the system.
We be must able to decide whether it is worth
storing the value in the first place.
Consider the case of multiplication -- it is
quicker to recompute the product of two
numbers rather than store a large
multiplication table.
How can we decide?

 Cost-benefit analysis -- Decide when the

information is first available whether it should be
stored. An analysis could weigh up amount of
storage required, cost of computation, likelihood
of recall.

 Selective forgetting -- here we allow the

information to be stored initially and decide later
if we retain it. Clearly the frequency of reuse is a
good measure. We could tag an object with its
time of last use. If the cache memory is full and
we wish to add a new item we remove the least
recently used object.
Learning by Taking Advice
There are two basic approaches to advice taking:
 Take high level, abstract advice and convert it
into rules that can guide performance elements of
the system.
 Develop sophisticated tools such as
knowledge base editors and debugging.
These are used to aid an expert to translate his
expertise into detailed rules.
Learning by Problem Solving
There are three basic methods in which a system can
learn from its own experiences.

1. Learning by Parameter Adjustment

The basic idea of parameter adjustment is to:
 Start with some estimate of the correct weight settings.
 Modify the weight in the program on the basis of
accumulated experiences.
 Here the problem has an evaluation function that is
represented as a polynomial of the form such as:

c1t1+c2t2+c3t3+…………
 The t terms a values of features and the c terms are
weights.
2. Learning by Macro Operators

The basic idea here is similar to Rote

Learning:
Avoid expensive recomputation
Macro-operators can be used to group a
whole series of actions into one.

3. Learning by Chunking

Chunking involves similar ideas to Macro
Operators and originates from psychological
ideas on memory and problem solving.
Weak Slot and Filler Structures

Why use this data structure?

It enables attribute values to be retrieved
quickly
assertions are indexed by the entities
binary predicates are indexed by first argument.
E.g. team(Mike-Hall , Cardiff).
Properties of relations are easy to describe .
It allows ease of consideration as it
embraces aspects of object oriented
programming.
A slot is an attribute value pair in its simplest
form.
A filler is a value that a slot can take -- could
be a numeric, string (or any data type) value
or a pointer to another slot.
We will study two types:
 Semantic Nets.
 Frames.
Semantic Nets
The major idea is that:
The meaning of a concept comes from its
relationship to other concepts, and that,
The information is stored by interconnecting
nodes with labeled arcs.
Representation in a Semantic
Net
The physical attributes of a person can be represented as in Fig.

Fig: A Semantic Network

These values can also be represented in logic as: isa(person, mammal),

instance(Mike-Hall, person) team(Mike-Hall, Cardiff)
 Frames

A frame is a collection of attributes or slots

and associated values that describe some real
world entity.

Frames can also be regarded as an extension to

Semantic nets. As tasks became more complex the
representation needs to be more structured. The
more structured the system it becomes more
beneficial to use frames.

Each frame represents:

 a class (set), or
 an instance (an element of a class).
Frame Knowledge Representation