Machine Translation, 7.5hp, VT 2016

Mid-course evaluation

This page contains a summary of the mid-course evaluation performed April 27, 2016. We received comments from 13 students, mixed with bachelors and masters. I have summarized all comments below. Comments marked with an asterisk were given by several students.

Lectures

  • Great *
  • Fabienne's lectures are great *
  • Good with a mix of teachers *
  • Basic and introductory
  • Good with exercises given by Fabienne
  • Too much reading, and more advanced than the lectures
  • Guest lectures felt unneccessary
  • Guest lectures could have been better placed in the course
Sara's comment:
We are happy to see that the majority are pleased with the lectures. The reading is there to give you pointers to more details and if you want to learn more. It is impossible for us to cover everything in the lectures, and we have to base them on the level of the prerequistie courses. I think the guest lectures are a good addition ot the course, to give you an insight into how MT can be used in industry, but we will look into how they fit in the course for next year.

Labs

  • Good *
  • Too basic
  • Time consuming
  • Challenging and a good level
  • Too many tasks
  • Graded labs would be better
  • The labs are too long
  • Lab 1: lots of menial labour
  • Lab 2:
    • Tedious
    • Good in a challenging way
    • Too many small questions
    • Too many small tasks
  • Lab 3: Slow and boring to download many files *
Sara's comment:
You have quite mixed impressions. It is hard to make labs that pleases everybody. There is a fair amount of work in the labs, but that is on purpose. You are expected to put around 200 hours of work into the course, and the labs + project/assignments + lectures and reading are meant to amount to around that workload. There is, however, always room for improvement, and we will take your comments into account when updating the labs for next year. Lab 2 was changed this year with the addition of the blind test set and automatic evaluation, which I think worked well, but I think other parts of the lab can be improved. We had graded labs in the past, but we think this system is better, especially since you now work in pair, in order for the grading to be fair.

Project

  • Good that groups were selected based on topic
  • What is the scope?
  • We haven't started yet *
Sara's comment:
I think you need to start on your project very soon. Your supervisor will help you settle on a scope when you meet up, but in general for the presentation you should give a theoretical overview of your topic, which in many cases could be based on the course books. For your project you should setup, run and evaluate MT systems for some aspect(s) related to your topic, and read some research articles in addition to the course book. The final report can be expected to be somewhere around 5-10 A4 pages.

Other

  • Slides online hard to find, webpage or protalen?
  • Assignments could be given earlier
  • Very good course
  • Better with slides before the lectures
  • Too many deadlines
  • I want to learn about translation for a specific language pair
Sara's comment:
The slides are available on the web page after each lecture. There is no material in portalen, but we store your progress there. We understand that it would be good to have slides earlier, but mostly we do not have them ready in time for that. There are many deadlines, but we have found that it is better to have a deadline for each lab, than just to have one deadline at the end of the course, since more students hand in their work with this system. We have made an effort to spread out the deadlines in this and the IR courses, since we thought that would help with your workload. I agree that it would have been great if the labs were available earlier. For the bachelor assignments we decided this year that 2 weeks was a good time period to complete assignment 1 and 3. We will evaluate this for next year, and maybe change it. It is possible to start work on assignment 2, however, which nearly no one has done yet. We will not discuss much about translation for specific languages, except a bit in the lecture about morphologically rich languages. If you are a master student you might be able to choose an article for your individual reflection report on a specific language, depending on your topic.