Scribd
Upload
10 views6 pages

KR NML

Non-monotonic reasoning addresses the limitations of conventional reasoning systems by allowing for conclusions based on incomplete or uncertain information. It includes methods such as default reasoning and abduction, which help manage conflicts and update knowledge bases as new facts emerge. The document discusses the principles and applications of non-monotonic logic, including how it differs from traditional logic in handling inconsistencies and assumptions.

Uploaded by

armaanrajco22a303
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
10 views6 pages

KR NML

Non-monotonic reasoning addresses the limitations of conventional reasoning systems by allowing for conclusions based on incomplete or uncertain information. It includes methods such as default reasoning and abduction, which help manage conflicts and update knowledge bases as new facts emerge. The document discusses the principles and applications of non-monotonic logic, including how it differs from traditional logic in handling inconsistencies and assumptions.

Uploaded by

armaanrajco22a303
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 6

NON-Monotonic Reasoning

Conventional reasoning system, such as first order


predicate Logic works well when information is

1. Complete or all facts necessary to spoke


problem are present or can be derived by
conventional method.
2. It is consistent-
3. the only way it can change is that new facts can
be added if they are consistent with already existing
facts – monotonicty.

In real words the knowledge is often changing,


uncertain and Fuzzy. In tackling this situation,
following key issues;

1. How can knowledge base be expanded. How


inferences are to be drawn on presence of
knowledge and Lack of knowledge .

2. How can knowledge base be updated. When


facts are added or deleted, proofs of so many
statements become invalid.
3. How can knowledge base be used to help
conflicts when there are several inconsistent
inferences.

There are two commonly used system to tackle above


kind of problems.

1. NON- Monotonic reasoning: in which


axioms and/or and inference rules are extended
to reason with incomplete informations. Here
state is either true or False or not believed to be
either of them.

2. Statistically Reasoning : Here a numeric


measures of certainty is assigned to each
statement rather than simply True or False with
statement.

In non –monotonic reasoning we draw conclusion


based on what is most Likely to be true. We assume all
those Facts to be true which are not proved to be untrue.
If you have no reason to believe some one did not
commit murder then he did not. This is called
DEFAULT reasoning. There are two ways:

NON – Monotonic Logic : Here we agument First


order predicate Logic with new operator M called
Model and is read as is consistent means undecidable.

2 V x, y Related(x , y) ^ MGet Along(x, y)→ will


defend (x, y) fact x get data y is consist with every then.

3 ^ P to be False does not imply P to be true. We draw


inferences in conventional way. If we come across
inconsistency then

V x : Republic(x) ^ M7 pacifist(x) → 7 pacifist(x)

V x : Quaker(x) ^ M pacifist(x) →pacifist(x)

Republic(Dick)
Quaker(Dick)
but taken together conflict.

=7pacifist(Dick) and Pacifist(Dick)


No conclusion about pacifist(Dick) can be made. To
resolve conflicts we take intersection of all theorem
proved.

Auto epistemic Logic(more-1985) solves some of these


problem

Default Logic : -Reiter 1980)

A:B
C

If A is provable and is consist to assume B then


conclusion C different between Default Logic and
NML.

I. In DFL the new rules of inferences are basis of


computation, using them we give plausible
extension of knowledge base.

II. They are rules of inferences at can not be


manipulated by other rules.
Abduction

V x A(x) →B(x)

V x measles (x) →spot(x)

But spot(x) does not necessary imply measely(x) but it


can be a best guess. Such type of default reasoning is
called Abduction. We work in direction which is most
Likely to be true. We can attach certain measures to this
uncertainties.

Inherence : we have already seen in baseball example if


a object belonging to some class has a attribute
explicitly mentioned otherwise it inherits that attribute
(if present) from a boarder class. How do we express in
Logic.

In default Logic

Baseball Player(x) : height(x , 6-1)


Height (x , 6-1)
To assert the fact that it does not inherit height from
adult male
Adult-Mali(x) : 7 baseball –Player(x) ^ height (x , 5-10)
Height (x , 5-10)

We may end up in

Adult male(x):
7baseballPlayer(x) ^7Midget(x)^7 Jockey(x) ^Height(x , 5-10)
Height(x , 5-10)

V x : Adult -male(x) ^7AB(x, aspect1) →height(x , 5-10)

V x : Baseball –Player(x)→AB(x, aspect)

V x : Midnight (x) → AB(x, aspect)

V x : Jockey (x) → AB(x, aspect)

:7 AB (x, y)
7 AB (x , y)

You might also like