Re: Would like to find study group for computer science comprehensive exams.

I do not believe this. I make a simple request for a study
group or a tutor and I trigger a flame war. If anyone of
you in this argument wants to help me pass automata
theory, please email me at eternalsquire AT hotmail DOT com


On Oct 23, 11:40 am, Patricia Shanahan <p...@xxxxxxx> wrote:
tc...@xxxxxxxxxxxxx wrote:
In article <ffj3sn$2gj...@xxxxxxxxxxxxxx>,
Patricia Shanahan <p...@xxxxxxx> wrote:
Why not write questions that require the student to demonstrate
understanding, and let us use whatever resources we want?

For a lot of courses, it's unfortunately fairly difficult to create problems
that "test understanding" that are still solvable by the average student in
a couple of hours. Say you're a professor and you come up with a few good
problems like that for your exam. The next year, your students will bring
answers to those questions with them to the exam, having inherited them from
the students who took the class with you last time. Copying the answers
doesn't test understanding, but how do you block that?

...

Change the question in a subtle way that looks rather like the previous
question.

If someone has memorized or brought with them (depending on closed or
open book) a series of answers to previous questions they will reproduce
the answer to the old question and get it wrong.

If they have memorized or brought with them (depending on closed or open
book) basic facts and formulas and really solve the exam problems, they
will get it right.

For computer science theory, I've seen problems made more difficult by
leaving to the student the choice of which of two contradictory
propositions to prove. For example, a question might define a language
L, and then ask the student "Prove either that L is NP-complete or that
L is not NP-complete.". Subtle changes in the language definition change
which proposition is provable.

Patricia


.

Relevant Pages