Machine Translation (MT)

Response to 2015/16 survey feedback.

I have taken note of the feedback from students in this coursesurvey, and have the following comments in response.

  • Assessment has been a consistent theme of the comments: you said that the coursework is too time-consuming relative to its worth, that the exam is difficult to study for, and that there is too much work overall for a 10-credit course. I completely agree with this, but making the coursework less time-consuming would be a disservice to you, because machine translation is an applied field that it best learned through experience. That is why I have changed the assessment structure for 2016-17: machine translation is now a 20-point course assessed entirely by coursework, and there are now four lab sessions for even more hands-on experience. You should expect the coursework to be even more challenging this year, but I hope you will learn a lot from it!
  • A second common theme in the comments is that the slides are not useful for revision. The slides are designed to provide visual explanations alongside the lecture, rather than bullet-point notes to be read aloud. I strongly advise you to revise from Philipp Koehn’s excellent textbook, which you can read online for free through the University Library. However, I realize that not everyone will do this, so I will work to record lectures and provide lecture notes alongside the slides this year.

Adam Lopez, October 2016