Material covered on Jan 18th
Introduction to: the course and grading; Course project;
Question answering; Link Grammar; wordnet.
Importance and significance of Knowledge representation.
Material covered in Jan 20th
A bit on wordnet; what we mean by complex questions;
importance of knowledge representation, a bit of history, what
we are looking for in a knowledge representation language.
Material covered in Jan 25th
English to logic form example, logical form to facts
example, running that example in link grammar.
Going over the slides of the travel example,
illustrating what kind of domain one shoudl look for.
Properties that a good KR language should have.
Illustrating Smodels GUI.
Materials covered on Jan 27th
Many examples of AnsProlog programs and
discussion of their intuitive meanings.
Homework 1 assigned.
Homework 1. Due Feb 3rd. -- postponed to Feb 8th
For the travel domain, write an algorithm of how to translate
the output of link grammar and use wordnet to get the
facts that we want. (See Deepthi Chidambaram's M.S. thesis,
in particular, Chapter 6. In page 51 there is an algorithm given.
I would like a similar -- perhaps more detailed -- algorithm.)
Material covered on Feb 1st.
predicates, functions, constants, variables, Herbrand base,
Herbrand universe, grounding of programs,
Semantics of AnsProlog programs, Iterative fixpoint semantics.
Gelfond-Lifschitz transformation. Answer set semantics. Small examples.
Homework 2. Due Feb 8th.
Write 10 AnsProlog programs (of length 1, 2, ..., 10)
and run it using Smodels. Explain why the answer sets
you got are correct.
Material covered on Feb 3rd.
Lots of exanples on Stratification and Splitting.
Material covered on Feb 8th.
Small programming modules in AnsProlog: there exists,
for all, various enumeration, choice, many smodels
Material covered on Feb 10th.
Refuse admission example, Nqueens problem.
Home work 3: due Feb 17th.
Write an Smodels program that solves the following puzzle.
One week five bachelors, Andy, Bill, Carl, Dave, and Eric agreed to
go out together to eat the 5 evening meals (Thai, Fish, Tacos, steak, pizza)
on Monday through Friday.
It was understood that Eric would miss Friday's meal because of an
out-of-town wedding at which he fervently hoped to catch the bride's
garter. Each bachelor served as the host at a restaurant of his
choice on a different night. Use the clues below to determine
which bachelor hosted the group each night and what food he selected.
Carl hosted the group on Wednesday.
The fellows ate at a Thai restaurant on Friday.
Bill, who detests fish, volunteered to be the first host.
Dave selected a steak house for the night before one of the
fellows hosted everyone at a raucous pizza parlor.
Material covered on Feb 15th, and 17th.
Luis: Go over the homework on Link grammar and travel module.
Introduction to Progol.
Nam: Go over other declarative problem solving examples.
Material covered on Feb 22nd.
Discussed class project. Intuition behind classical
negation. Formalizing normally, exceptions and weak exceptions.
Material covered on Feb 24th.
Go over classical negation with lots of examples.
Michael's example on exceptions and weak exceptions.
Discuss project proposals.
Homework 4. Due March 1st.
The Detroit Red Wings won their 8th Stanley Cup, the championship of the
National Hockey League, in June 1997. During the off-season, the team
changed due to trades and injuries. The Wings' locker room manager is a
logic puzzle fan, and he decided to give 6 of the players their new
locker assignments in the form of a logic problem. The players have
different favorite away-from-home arenas around the league, and each
has a different favorite drink which he uses to slake thirst during a game.
After a game the players all have different favorite soaps with which
they shower. Each likes to relax after a game watching a different
genre of movie while eating a different favorite meal. For those who
wonder which Wings have which favorites, use the clues and the table
below to place each player in the correct locker and to determine his
favorite arena, drink, soap, meal, and movie genre.
One player's favorite away-from-home arena is New Jersey's Byrne Meadowlands;
one's favorite game drink is Jolt -- double the sugar, double the caffeine;
one's favorite soap is Dial;
one's favorite post-game meal is fried chicken;
one player likes to watch classical movies best.
Kris Draper, who likes Chinese food after a game, uses a locker next
to Slava Kozlov.
Goalie Chris Osgood's locker is next to that of the player who
enjoys pizza with a western after a game.
The locker numbers of the Sprite drinker and the skater who is
partial to adventure movies differ by 2.
The Coke drinker, whose favorite visiting rink is Colorado's McNichols
Arena, and the player whose favorite visiting rink is Toronto's Maple
Leaf Gardens, both have a locker next to the same player; this player
does not like Sprite.
Steve Yzerman, whose favorite visiting rink is the Kiel Center in St. Louis,
has a locker next to the Safeguard sudser.
The one who showers with Irish Spring does not like Calgary's Saddle Dome.
The player whose game drink is orange juice enjoys a western after a rough game.
His locker is next to the player who relaxes with horror movies.
Kozlov lockers next to the player who is partial to Anaheim's Arrowhead Pond.
The player who lathers with Lifebuoy has a locker next to both Darren McCarty
and the player who scrubs with Zest and guzzles Gatorade.
Osgood, the spaghetti eater, and the comedy movie fan (in some order) have 3
consecutive locker numbers.
The 2 players whose favorite visiting arenas are the Saddle Dome and the
Kiel Center each have a locker next to the skater whose game drink is water.
The lockers of McCarty and the player who favors Arrowhead Pond are not
adjacent, but there is only 1 locker between them.
Igor Larionov and the player who enjoys Coast cascades are locker neighbors.
Osgood, the Nachos muncher, has a higher numbered locker than the player
who loves steak.
The Coke drinker and the mystery movie fan have adjacent lockers.
Igor lockers next to the horror movie fan.
Both Kris Draper and the Coast soaper locker next to the steak eater.
Material covered on March 1st.
Go over project. Reasoning about actions: projection, planning,
explanation, and reasoning with incomplete information.
Homework 5: due March 3rd 2005
Consider the actions shoot and load. Load causes the gun
to be loaded and shoot makes the turkey dead (not alive) if the gun is loaded.
Now formulate the following reasoning using AnsProlog/Smodels.
1. Projection: Assuming that the gun was initially not loaded and the turkey
was alive project onto the future
what happens if (a) shoot happens in time point 0.
(b) load happens in time point 0 and shoot happens in time point 1.
2. Planning: Assume that the gun was initially not loaded and the turkey
was alive. Find a plan so that the turkey is dead in time point 2.
Assume that the turkey was initially alive; shoot occured in
time point 0 and the turkey was found to be not alive in time point one.
How can one explain this, in terms of whether the gun was loaded
or not initially.
4. Reasoning with incomplete information:
Assume that the turkey is initially alive and we do not know
if the gun is initially loaded or not. What can we say about time point 1
if shoot occured in time point 0.
Material covered on March 3rd.
Mostly going over the material covered on March 1st; situation
calculus notation; actions with durations; role of normative
reasoning, weak and strong exceptions in the context of
reasoning about actions.
Material covered on March 8th.
Explain homework 6. Graph colorability, Knapsack; Combinatorial
Material covered on March 10th.
Constrained LP in Sicstus.
March 14th -- 18th Spring break
March 22nd -- Constrained LP, Overview for the test.
March 24th -- Test 1.
Introduction to probabilities, Bayes rule,
joint probability distribution,
Intro to Bayes nets.
Pearl's functional causal models.
Homework 8: Due April 19th
19.4 and 19.5 (pages 340-341) from the handout (Chapter 19)
and compute P(Q | P8, P11, P2) with respect to figure 19.4 (page 333).
The Rifleman exanple.
P-log: Integrating logic and probability.
Rifleman example using P-log.
Learning Bayes nets.
MDPs and POMDPs.