Knowledge Representation and Reasoning 2012-2013
G53KRR module description
Last year web page
Previous exam papers and answers
Module feedback (average exam marks etc.)
Feedback for exam 2012-13
The module describes how knowledge can be represented symbolically and
manipulated in an automated way by reasoning programs. It will be based
on the following textbook:
Ronald Brachman and Hector Levesque. Knowledge Representation and
Reasoning. Elsevier, 2004.
There are several copies in the Jubilee Library, location Q387 BRA.
An a link to eBook edition from the library catalogue.
Hector Levesque's on-line lecture slides based on the book
local copy
Lectures
- Lecture 1 (1 Oct) (Introduction, what are knowledge based systems)
- Lecture 2 (4 Oct): syntax and semantics of first-order logic. See Levesque's slides 18-27 ,
textbook chapter 2.1 - 2.3.2.
- Lecture 3 (8 Oct) : semantics of first-order logic, logical entailment.
See Levesque's slides 28-34, textbook
the rest of chapter 2.
- Lecture 4 (11 Oct): expressing knowledge in first-order logic.
See Levesque's slides 35-45 , textbook chapter 3.
- Lecture 5 (15 Oct): Exercises:
Barber exercise
Answer to the barber exercise
Alpine Club exercise
Answer to the Alpine Club exercise
- Lecture 6 (18 Oct): answer to Alpine Club exercise and begin Resolution (textbook
chapter 4, Levesque slides from slide 46).
- Lecture 7: propositional resolution, chapter 4.1, slides 46-55.
Handout with definitions and rules
Exercise on propositional reduction to CNF
Answer to the exercise
Exercise on propositional resolution
Answer to the exercise
- Lecture 8: Resolution continued. First-order case. Slides 56-61.
Handout with definitions and rules
Exercise on first-order resolution.
Another exercise on first-order resolution: answer Alpine club exercise using resolution
Solution to both exercises
Exercise on answer extraction
Solution to the exercise
- Lecture 9: continue first-order resolution (Skolemisation, answer extraction...). Slides 62 and 64.
- Lecture 10: Unification. Algorithm for computing a most general unifier. Handling equality. Slides 64, 68. Handout about unification and equality
- 5 November: Lecture 11: Termination and complexity of resolution. Slides
66, 67, 71, 72.
- 8 November: Lecture 12: Horn clauses. SLD resolution. Backward chaining. Chapter 5, slides 79-90. Exercise: do question 4 in 2010-2011 and 2011-2012 exam.
- 12 November: Lecture 13: Prolog. SEM feedback.. Reading: Chapter 6, slides 91-100. If you want to try Prolog,
it is installed on clyde, type sicstus at command prompt. User manual is
here .
- 15 November: Lecture 14: Horn clauses continued. Forward chaining.
Handout about forward chaining and exercise
Rules in production systems. Chapter 7, slides 89, 103-118.
Answer to the exercise
- 19 November: Lecture 15: Life cycle of a knowledge based system. Knowledge acquisition.
Decision tables. Java Rules API and Jess.
Lecture slides
There is a small exercise on decision tables in the lecture slides, and
you can do exam 2007 (5b), 2008 (4d), 2009 (6b), 2011 (5a).
- 22 November: Lecture 16: Description logic. Chapter 9, slides 138-163.
Summary of technical definitions and an exercise
Answer to the exercise
See also
exam papers
2010: question 3, 2009: question 4, 2008: question 5.
- 26 November: Lecture 17: Non-monotonic reasoning. Chapter 11, slides 178-201.
- 29 November: Lecture 18: Non-monotonic reasoning continued.
Summary of lectures 17 and 18
See also
exam papers
2010: question 5, 2009: question 5, 2008: question 6.
- 3 December: Lecture 19: Bayesian networks. Chapter 12, slides 202-216.
Summary of Bayesian networks and an exercise
(There was a typo in the handout, Pr( a UNION b) = Pr(a) + Pr(b) -
Pr( a UNION b) instead of Pr( a UNION b) = Pr(a) + Pr(b) - Pr( a INTERSECTION b), fixed now, thanks to Laurence Herbert.
Solution to the exercise
- 6 December: Revision.
- 11 January: Revision 2.
This file is maintained by Natasha Alechina
Last updated 19 February 2013.