Chapter 10 - Final

Question 1

In more complex domains, how do we deal with representations?

Answer 1

We concentrate on general concepts:
Events
Time
Physical objects
Beliefs

These are abstract concepts that are very general and thus useful.

Question 2

What is ontological engineering?

Answer 2

Ontological engieering is referred to as focusing on general abstract concepts of representing knowledge with:
Events
Time
Physical objects
Beliefs

Question 3

We cant represent everything in the world. How can we deal with this?

Answer 3

we use placeholders that allow new, important and relevant knowledge to be stored at a later time. By creating some sort of outline or interface for knowledge, we’re building a system that can be implemented at a later stage.

We can do this by for instance describing what it generally means to be a physical object, what it means to be a book etc. The further details can be specified at a later time.

Question 4

name an example of a “placeholder” for something

Answer 4

JavaFX window. The window is a general concept with a certain behavior. It is up to the user to implement the specifics of it.

The user can utilize the window, and create a more specialized object, like a SpreadSheet window.

Question 5

What is upper ontology?

Answer 5

Upper ontology refers to the general framework of concepts.

Question 6

What are the characteristics of a general purpose ontology?

Answer 6

1) A general purpose ontology should be applicable in any special purpose domain.

2) In sufficiently demanding domains, different areas of knowledge has to be unified. This is because in such domains, we often use knowledge from different “concepts” together. For instance, driving a car requires information and knowledge about time, objects, behavior of humans etc.

Question 7

Is general purpose ontology a success?

Answer 7

No really. It is most common to use special purpose knowledge engineering and machine learning.

Question 8

Elaborate on the use of categories in knowledge representation

Answer 8

Categories are vital. This is because most interaction with “things” happens at category level. In many cases, knowledge of the category is enough for the purpose.

Inferring category membership is one thing that is extremely common and important.

Question 9

What choices do we have in representing categories in FOL?

Answer 9

1) Predicates
2) objects

We could for instance use the predicate Basketball(b) to indicate membership. OR, we could create an object, called Basketballs, that represent a category. Then we’d say BB9 is a member of, or subset of, Basketballs.

Question 10

Do categories use inheritance?

Answer 10

Yes

Question 11

What is taxonomy?

Answer 11

Taxonomy refers to a hierarchy of groups that have been named and placed so that they form some sort of relation.

Question 12

What are disjoint categories?

Answer 12

Categories are disjoint if they share no members in common

Question 13

What is an exhaustive decomposition?

Answer 13

Includes all objects of some type.

An exhaustive decomposition of disjoint sets is a partition.

Question 14

How do we generally say that some objects is a part of something else? like, a page is a part of a book?

Answer 14

we use the PartOf relation.

PartOf(Romania, Europe)

PartOf(Nose, Person)

Question 15

Can we assign properties to a set of objects?

Answer 15

NO. this would not make any sense, as a set of objects is nothing but a mathematical abstraction.

We need to define some object or relation that can hold properties. For instance, a set of apples can be described as a “bunch”.

BunchOf({Apple1, Apple2, Apple3})

BunchOf creates a composite object. Not to be mistaken with “Apples”, which would be a category.

Question 16

What is a composite object?

Answer 16

A composite object is an object that consists of other object. NOT TO BE MISTAKEN WITH CATEGORIES!

A composite object could be a human, as a human consists of legs, eyes etc.

Generally, we use BunchOf(set) to create a composite object.

Question 17

What is important regarding natural kinds of cateogires?

Answer 17

They have no exact definition.

Question 18

Elaborate on measures

Answer 18

We use measure objects.

For instance, we can define a “length object” of 1.5 inches. We then represent length with a “units function” that takes a number as argument.

Diameter(Basketball) = Inches(9.5)

Weight(BunchOf({Apple1, Apple2})) = Pounds(2)

Question 19

Regarding measures, how do we deal with more complex cases where an exact measure is difficult? For instance, difficulty or beauty.

Answer 19

There is no point in trying to place a number on it.

Instead, we use relative ordering.

By finding out which exercises are more difficult, one can use this to make decisions. For instance, avoid difficult ones.

Question 20

What is qualitative phsyics?

Answer 20

A subfield in AI that focus on how to reason about physical systems without using detailed equations and numerical simulations.

Question 21

Elaborate on things vs stuff

Answer 21

it is a matter of individuation. Some objects are composite, consisting of differnet objects. For instance, a car consists of other obvious individual objects.

However, butter consists of butter, and we can split it, which gives us 2 items with the exact same behaviour, but smaller. We call such items “stuff”.

We dont say “one butter” unless we’re talking about a specific bottle or box of butter.

Formally, we can make the distinction between count nouns and mass nouns

Question 22

intrinsic vs extrinsic properties

Answer 22

Stuff vs things. Intrinsic values belong to the substance intself.

Extrinsic values belong to the object.

Question 23

Recall fluents

Answer 23

Fluents are aspects of the world that can change

Question 24

Discuss briefly continous vs discrete time considerations

Answer 24

It is simple if we treat time as “fixed” and instant when things happens. for instance, if we fill a bath tub, and we treat this action as “instant”, we’re unable to do anything during the filling time.

By recognizing that a lot of time events has a duration, things become more complicated.

We use “event calculus” to treat such time cases

Question 25

Elaborate on event calculus

Answer 25

event calculus use the following objects:
1) Events
2) Fluents
3) Time points

A fluent could be: At(Shankar, Berkeley) represent Shankar being at Berkeley at this time.

An event: E_1 “e” Flyings AND Flyer(E_1, Shankar) AND Origin(E_1, SF) AND Destination(E_1, DC)

This event means: Event 1 is included in the Flyings-category, and Shankar is the flyer, and the event is a flying from SanFran to DC.

By specifying events like this, we can add any arbitrary information.

To assert that a fluent is actually true at some time t1 and continuing to t2, we use the following:
T(At(Shankar, Berkeley), t1, t2)

More generally: T(f, t1, t2). This says that fluent f is true between t1 and t2.

Question 26

List the complete set of predicates for one version of event calculus

Answer 26

T(f,t1,t2): Fluent f is true between t1 and t2.

Happens(e, t1, t2): Event e starts at time t1 and ends at time t2

Initiates(e, f, t): Event e causes fluent f to become to at time t.

Terminates(e, f, t): Event e cause fluent f to be FALSE at time t.

Initiated(f, t1, t2): Fluent f becomes true at SOME point between t1 and t2.

Terminated(f, t1, t2): Fluent f cease to be true at some time between t1 and t2.

t1 < t2: Time point t1 occurs before time point t2.

Question 27

Elaborate on time in event calculus

Answer 27

We consider 2 types of time:
1) Time intervals
2) Moments

The key is that moments have 0 (zero) duration.El

Question 28

elaborate on mental objects

Answer 28

This is about understanding what sort of questions we are asking, and what knowledge is used. For instance, some questions are impossible to answer, and by knowing this, we can be more efficient. As an example, it iwll not help to “think harder” if the question is, is the president sitting down or not.