CPSC 322 - Lecture 22

Knowledge Representation

I.  Knowledge representation

We've spent a lot of time talking about search, because
when you get right down to it, that's the backbone of
lots of work in artificial intelligence.  But there's 
more to AI than just search procedures.  Walking hand
in hand with search, heading down the road in the
general direction of intelligent behavior, goes the issue
of knowledge representation.  No matter how good your
search procedures are, if you don't make the right 
choices with respect to knowledge representation, your
"intelligent" program is going to take a long long time
to find the answers you were hoping for, if it finds them
at all.

In this lecture we presented a few desirable attributes of
a generic knowledge representation scheme, and we made
the connection between building AI systems and building
any kind of large software system.  At some level, they
both can be viewed as problems in software engineering,
with developers asking and answering the same sorts of
questions about process and representation regardless of
whether they're building AI systems or some other 
applications.

We began exploring some of those questions, and possible
answers, in the context of a simple example in language
understanding.