| |
COMP348 Document Processing and the Semantic Web
Assignment 1, Part 1:
Feedback
This was generally done well, as I'd hope for the introductory assignment. I have only a few
general comments:
-
As discussed in lectures, a suitable baseline was just the accuracy of choosing the most frequent
category. (You could have chosen another baseline if you had a good argument for it, but noone really
did.) Most people did OK here. However, in some cases there was a "I chose 80% because I felt it was
a good baseline". Baselines aren't chosen on the basis of feelings; they're chosen on the basis of data.
If you don't have the hang of the concept yet, ask me.
-
Make sure you include all the relevant numbers in the report. Don't just give me the formula for
the z-test of proportions and expect me to plug in the numbers. It's a report, so you're supposed
to be reporting things like that.
-
With the z-score calculation, a few people reported it to 10 decimal places. You probably only want
to report it to 2 d.p.; more decimal places suggests spurious precision. (Wikipedia has a bit of a discussion about
this.)
-
Presentation didn't matter here, but will in Part 2. You want it to look like an actual report (albeit
shorter) of the sort you'll have to write when you go and work.
There will be more details in Part 2, but
here are some comments now: the report should have a title; it should be typo-free; it should be in
satisfactory English; it should have a decent layout (tables can be nice).
Mark Dras or
|
