CSCI-8050: KNOWLEDGE BASED SYSTEMS - FINAL EXAM
Note: Prepare solutions using your favorite word processor. Hand in your solutions (right, only a
hardcopy is acceptable) on or before the day of the final exam, no later than
1) (20) Why is multiple fault diagnosis an appropriate domain for
applying expert systems? Be very
specific and complete. Include examples
of major expert systems to support your answer if there are any.
2) (20) What are the differences between deep and shallow expert
systems? Be sure to include support for
and examples in your answer.
3) (20) What are the basic components of a production system and
what components do they correspond to in an expert system architecture? Explain.
4) (20) Identify 4 domains where using a production system would
be considered inappropriate, and 4 where it would be considered
appropriate. Explain your answers.
5) (10) Develop a set of Newell’s PS style production rules to “split”
a positive (> 0) binary integer. That
is, given a positive binary integer in the database, apply rules so that the
resulting value left in the database is one-half the original value (just
truncate any remainders). For example,
given 1100, we apply our set of rules and end up with 110. Be sure to describe your recognize/act cycle
and any characteristics necessary for proper use.
6) (10) Now, develop another set of Newell’s PS style production
rules to “double” a positive binary integer.
7) (10) Compare/contrast one of: (case-based reasoning systems or
truth maintenance systems) with rule-based systems. Identify at least 4 major issues for your
comparison. Be sure to support your
answer with citations and in-depth analysis (not just definitional issues).
8) (20) Given that: “Bob tells the truth 90% of the time.”, “Jim
tells the truth 75% of the time.”, and “A fair six-sided die has been tossed.”
determine the following:
a)
what
is the probability that the die came up a 6?
b)
what is
the probability that the die came up a 6 given that Bob said it came up a 6?
c)
what
is the probability that the die came up a 6 given that Jim said it came up a 6?
d)
what
is the probability that the die came up a 6 given that Jim said Bob said it
came up a 6?
9) (20) Discuss the computational power of production systems and
then compare/contrast this power with the computational power of Turing
Machines.
10) (30) Present a short overview of MYCIN's approach to dealing
with uncertainty (i.e., the use of certainty factors which are based on belief
and disbelief measures; in the pre-1977 approach each hypothesis had both an MB
and MD stored instead of a single CF).
Compare/contrast this approach to reasoning under uncertainty with one
of: (Dempster-Schafer Theory of Belief Functions or Fuzzy Logic). Focus on 4 issues that you consider the most
important.
11) (20) "One of the most often heard criticisms of AI is
that machines cannot be called intelligent until they are able to learn to do
new things and adapt to new situations, ..." Argue either for or against this statement
using expert systems as your platform.
12) (10) What are the differences between an expert system and a
knowledge-based system? Explain, and use
examples to support your answer.
13) (30) Design and discuss
an intelligent information system for handling venomous snakes. Be sure to include features for handling
snake characteristics, anti-venom, snake identification, bite diagnosis,
treatment, patient prognosis, and whatever else you feel would be crucial to a
useful system.
14) (20) What major stages or steps would you recommend to someone
building a KBS? In which stages would
interaction with the expert be most crucial?
Does this change with varying approaches to acquiring knowledge from the
expert?
15) (20) Given boolean variables p, q, r, s, & t and the
expression
[(p or q) => r] => [(s and t)
=> r]. where => stands for
'implication'
There is exactly one combination of variable value assignments
that results in the expression's value being FALSE. What are these values? What is the total number of variable value
assignment combinations? Characterize
your problem solving strategy on this problem.
Now, comment on the complexity of automating a strategy (not necessarily
the one you used) to solve this type of problem (include guaranteed methods,
heuristic approaches, and search space size and growth characteristics).
16) My signature below attests to the fact that I neither gave nor
received any unauthorized assistance in the preparation of the answers to this
exam. All answers are my own work and
all supporting material has been properly referenced and cited; all in
accordance with the university and class policy on academic honesty.
Signature:________________________________________ Date: _________________