FUI's course evaluation

INF5820 - Språkteknologiske anvendelser

Assessment:

Choose a question to see data from the student assessment:
How do you rate the course in general? How do you rate the level of the course? How do you rate the work load in proportion to the number of credits achieved? The quality of the lectures? The lectures' relevance and usefulness in regard to relaying what you need to learn? The quality of the group tuition as a whole? In what degree does group tuition and weekly assignments give you practical skills and the possibility to make use of the theory at hand? Are the curriculum and learning outcomes of the course sufficiently defined? How do you rate the course material used (text book etc.)? How do you rate the difficulty level of the obligatory assignments? What is your opinion in regard to the number of obligatory assignments? How do you rate the difficulty level of the final examination? Show all Hide

Summary:

The respondents in general find this course to be Good or Very good. Most students rate the difficulty as OK or Difficult. Overall the respondents say that the workload is OK. The quality of the lectures is regarded as Very good to Good, and their relevance and usefulness as Good. Some students feel there should be more Group tuition on blackboard and Weekly assignments. The difficulty level and number of obligatory assignments is rated OK, while the difficulty level of final examination is considered High. The online course presentation is rated in general as being Exceptionally good or Very good, with respondents in general being able to find the information they needed.

From the text responses it seems that a few students are critical of the teaching material used for the part of the course concerning distributional semantics. They consider it to be too much reading, and the information hard to understand.

Quotes:

In my opinion the second part of the course (distributional semantics) was much more difficult, but not because of the subject itself, but because of teaching material used. All the scientific papers we had to read were difficult to understand and I had to read them several times to get the information I needed out of it. Besides, about 2/3 of the lectures were about machine translation and 1/3 about distributional semantics, while on the exam it was 50-50.

The course suffers from two problems. First, in the MT part, the textbook (Koehn) is not very good. It attempts to present the information in a easy-to-follow manner, but ends up being very hard to follow, on the border to confusing. The part on MT in Jurafsky and Martin is much better, but perhaps a bit lacking in depth. The second problem was in the part on distributional semantics. The curriculum was ill defined with too much reading considering it was only half the course. The reading list was published on the same day as the lecture making it hard to come prepared to class. Additionally, some the questions given on the exam were not really covered in the curriculum and only briefly mentioned in a lecture, while others were asking for very specific details of a specific article thus rewarding verbatim reproduction of the curriculum as opposed to testing general understanding of core concepts.

About the course evaluation

The course evaluation is done by students in FUI (Fagutvalget ved Institutt for informatikk), and is based on questionnaires made at the end of each semester.