Principles ofKnowledge Representation and Reasoning

Albert-Ludwigs-Universität Freiburg

Bernhard Nebel, Stefan Wölfl, and Julien HuéWinter Semester 2012/2013

Lecturers

Prof. Dr. Bernhard Nebel Room 52-00-028Consultation: Wed 13:00-14:00 and by appointmentPhone: 0761/203-8221email: nebel@informatik.uni-freiburg.de

Dr. Stefan Wölfl Room 52-00-043Consultation: by appointmentPhone: 0761/203-8228email: woelfl@informatik.uni-freiburg.de

Dr. Julien Hué Room 52-00-041Consultation: by appointmentPhone: 0761/203-8234email: hue@informatik.uni-freiburg.de

Nebel, Wölfl, Hué – KRR 2 / 22

Lectures

WhereLecture hall, Geb. 52, SR 02-017

WhenWed: 08:00–10:00, Fri: 08:00–09:00 (+ exercises)

Web pagehttp://www.informatik.uni-freiburg.de/˜ki/teaching/ws1213/krr/

Nebel, Wölfl, Hué – KRR 3 / 22

Exercises

WhereLecture hall, Geb. 52, SR 02-017

WhenFri, 09:00-10:00

Exercise assistant: Matthias WestphalRoom 52-00-041, Phone: 0761/203-8229email: westpham@informatik.uni-freiburg.de

Nebel, Wölfl, Hué – KRR 4 / 22

Exercises II

Exercises will be handed out and posted on the web pagethe day of the lecture.Solutions can be given in English and German.Students can work in pairs and hand in one solution.Larger groups and copied results will not be accepted.Previous week’s exercises have to be handed in before thelecture.

Nebel, Wölfl, Hué – KRR 5 / 22

Examination

An oral or written examination takes place in the semesterbreak.The examination is obligatory for all Bachelor/Master/ACSMaster students.

Admission to the exam: necessary to have reached at least50% of the points on exercises and projects.

Nebel, Wölfl, Hué – KRR 6 / 22

Course prerequisites & goals

GoalsAcquiring skills in representing knowledgeUnderstanding the principles behind different knowledgerepresentation techniquesBeing able to read and understand research literature in thearea of KR&RBeing able to complete a project in this research area

PrerequisitesBasic knowledge in the area of AIBasic knowledge in formal logicBasic knowledge in theoretical computer science

Nebel, Wölfl, Hué – KRR 7 / 22

AI and Knowledge Representation

AI can be described as: The study of intelligent behaviorachieved through computational meansKnowledge representation and reasoning could then beviewed as the study of how to reason (compute) withknowledge in order to decide what to do.

Knowledgeacquisition

Knowledgereasoning

DecisionAction

Nebel, Wölfl, Hué – KRR 8 / 22

Knowledge

We understand by “knowledge” all kinds of facts about theworld.It is more than just data. It is data+meaning.Knowledge is necessary for intelligent behavior (humanbeings, robots).

Nebel, Wölfl, Hué – KRR 9 / 22

Representation

If A represents B, then A stands for B and is usually moreeasily accessible than B.As those are surrogates, imperfection cannot be avoided.In our case we are interested in groups of symbols thatstand for some proposition.

Knowledge RepresentationThe field of study concerned with representations of propositions(that are believed by some agent).

Nebel, Wölfl, Hué – KRR 10 / 22

Reasoning

Reasoning is the use of representations of propositions inorder to derive new ones.While propositions are abstract objects, theirrepresentations are concrete objects and can be easilymanipulated.Reasoning can be as easy as arithmetics mechanicalsymbol manipulation.For example:

raining is trueIF raining is true THEN wet street is truewet street is true

Nebel, Wölfl, Hué – KRR 11 / 22

Why is Knowledge Representation andReasoning useful?

Describing/understanding the behavior of systems in termsof the knowledge it has.Generating the behavior of a system!

Declarative knowledge can be separated from its possibleusages (compare: procedural knowledge).Understanding the behavior of an intelligent system in termsof the represented knowledge makes debugging andunderstanding much easier.Modifications and extensions are also much easier toperform.

Nebel, Wölfl, Hué – KRR 12 / 22

Deduction/abduction/induction

A CB

A reasoning process usually consists in 2 out of 3 parts:antecedant, inference rule and conclusion where the objective isto find the third one.

Conclusion is missing: deductionInference is missing: inductionAntecedant is missing: abduction

Nebel, Wölfl, Hué – KRR 13 / 22

Deduction/abduction/inductionInduction

A CB

Inductiondatamining, economy

ExampleCase: These beans are [randomly selected] from this bag.Result: These beans are white.Rule: All the beans from this bag are white.

Example from Charles Sanders Peirce

Nebel, Wölfl, Hué – KRR 14 / 22

Deduction/abduction/inductionAbduction

A CB

Abductionmedical diagnosis, car repairing, failure explanation

ExampleRule: All the beans from this bag are white.Result: These beans [oddly] are white.Case: These beans are from this bag.

Example from Charles Sanders Peirce

Nebel, Wölfl, Hué – KRR 15 / 22

Deduction/abduction/inductionDeduction

A CB

Deductionmathematics

ExampleRule: All the beans from this bag are white.Case: These beans are from this bag.Result: These beans are white.

Example from Charles Sanders Peirce

Nebel, Wölfl, Hué – KRR 16 / 22

Deduction/abduction/inductionUnsound deduction

A CB

Deductioncommon-sense reasoning

ExampleUsually all birds fly.Tweety is a bird.Then Tweety normally flies.

Nebel, Wölfl, Hué – KRR 17 / 22

The role of complexity theory (1)

Intelligent behavior is based on a vast amount ofknowledge.Because of the huge amount of knowledge we haverepresented, reasoning should be easy in the complexitytheory sense.Reasoning should scale well: we need efficient reasoningalgorithms.

Nebel, Wölfl, Hué – KRR 18 / 22

The role of complexity theory (2)

Use complexity theory and recursion theory to

determine the complexity of reasoning problems,compare and classify different approaches based oncomplexity results,identify easy (polynomial-time) special cases,use heuristics/approximations for provably hard problems,andchoose among different approaches.

Nebel, Wölfl, Hué – KRR 19 / 22

Course outline

1 Introduction2 Reminder: Classical Logic3 A New Logic: Boxes and Diamonds4 Quantitative vs Qualitative logics5 Nonmonotonic Logics : Default logic and ASP6 Cumulative logics7 Belief change8 Description Logics9 Qualitative Spatial and Temporal Reasoning

Nebel, Wölfl, Hué – KRR 20 / 22

Literature I

R. J. Brachman and Hector J. Levesque,Knowledge Representation and Reasoning,Morgan Kaufman, 2004.

C. Beierle and G. Kern-Isberner,Methoden wissensbasierter Systeme,Vieweg, 2000.

G. Brewka, ed.,Principles of Knowledge Representation,CSLI Publications, 1996.G. Lakemeyer and B. Nebel (eds.),Foundations of Knowledge Representation and Reasoning,Springer-Verlag, 1994

W. Bibel,Wissensrepräsentation und Inferenz,Vieweg, 1993

Nebel, Wölfl, Hué – KRR 21 / 22

Literature II

R. J. Brachman and Hector J. Levesque (eds.),Readings in Knowledge Representation,Morgan Kaufmann, 1985.

B. Nebel,Logics for Knowledge Representation,in: N. J. Smelser and P. B. Baltes (eds.), International Encyclopedia ofthe Social and Behavioral Sciences, Kluwer, Dordrecht, 2001.B. Nebel,Artificial Intelligence: A Computational Perspective,in: G. Brewka, ed., Principles of Knowledge Representation, Studiesin Logic, Language and Information, CSLI Publications, 1996,237-266.

Proceedings of the International Conference on Principles ofKnowledge Representation and Reasoning,(1989, 1991, 1992, . . . , 2004, 2006), Morgan Kaufmann Publishers.

Nebel, Wölfl, Hué – KRR 22 / 22