Course development report 2025

This is the 2025 course development report for TDDE09 Natural Language Processing (6 credits).

Statistics

The 2024 session had 69 registered participants. Based on social security numbers, 62 out of these were men and 7 were women. There were 3 exchange students (France, Chile). The examiner was me, Marco Kuhlmann.

After the first examination (2024-03-23), 43 out of 69 students (62%) had passed the course. This number had increased to 57 out of 69 students (83%) after the last examination (2024-08-31).

Course evaluation

The Evaliuate for the 2024 session was open between 2024-03-11 and 2024-04-07. It received responses from 23 out of 69 respondents (response rate: 33%). The overall grade for the course was 3.87 out of 5 (median: 4.00).

Below is a summary of the free-text replies.¹

What students like about the course

Engaging and High-Quality Content: Students appreciated the course content, describing it as highly relevant, interesting, and insightful. Many found the lectures and labs effective in building a deeper understanding.

Teaching and Support: The teaching staff, including the examiner and TAs, received praise for their knowledge, helpfulness, and enthusiasm. Students felt supported when help was available.

Flipped Classroom and Quizzes: Several students found the flipped classroom approach combined with quizzes to be helpful for processing new information. The ability to re-watch recorded lectures at their own pace was highlighted as a significant benefit.

What students think can be improved

Workload and Scheduling: Many students felt the workload was too high, with frequent deadlines and limited time for completing assignments. They suggested spreading out the labs and project work more evenly to reduce stress, especially during the exam period.

Lab Assistance and Resources: The availability of teaching assistants (TAs) during lab sessions was a common issue. Students reported long wait times for help and suggested hiring more TAs or increasing lab support to improve the experience.

Clarity in Instructions and Grading: Students noted confusion around grading systems, project expectations, and instructions for assignments like the post-project paper. They recommended clearer and more consistent guidance to avoid misunderstandings.

Examiner comments

I want to thank all students who participated in the Evaliuate and took the time to share their thoughts. Thanks also to everyone who contributed to the separate evaluation conducted by LinTek (via D-sektionen). I really appreciate your feedback!

Unfortunately, the results from Evaliuate were not what I had hoped for. While many students expressed their appreciation for the course, the responses also revealed that I was unable to solve the biggest problem from previous years: reducing the perceived workload. On the contrary, 83% of the respondents found the workload too high, compared to only 47% last year. This highlights the need for significant changes (especially with the labs) to create a more manageable workload.

Another thing I am concerned about in the Evaliuate is the strong polarisation in the responses: While the majority of respondents (18 out of 23, or 78%) gave the course an overall grade of 4 or 5, a smaller but significant portion (5 out of 23, or 22%) gave the course the lowest possible grades of 1 or 2. Unfortunately, the free-text responses do not provide enough detail for me to understand what exactly went wrong for these students.

I have tried to incorporate student feedback into the course revisions for the 2025 session. Here is how:

Redesign of the lab format. I have thoroughly reworked the lab series in an ambition to make it more manageable. In previous years, the labs involved complex coding problems, and students spent a significant amount of time debugging, which often distracted from learning the core concepts, models, and algorithms. The new labs emphasise explaining and modifying existing code rather than starting from scratch. Additionally, I have broken the problems into smaller, more manageable chunks, hopefully making it easier for students to track their progress and feel a sense of accomplishment.

Redesign of the lab examination. I have also made significant changes to the lab module’s examination format. Previously, students had to submit one lab per week, a format they criticised for creating constant pressure. This criticism is in line with the pedagogical literature (Lauvås and Jönsson, 2019), which also warns that continuous examination shifts focus away from meaningful learning toward simply staying on top of deadlines. This year, I am introducing a portfolio-based assessment. Students will use the labs as preparation for a final oral exam, which I hope will reduce stress and encourage deeper engagement with the material.

Changes to the project. To address concerns about the workload in the project module, I have reduced the number of required project deliverables from seven to five. Additionally, I have clarified the structure of the post-project paper to make expectations for each section more transparent. For the first section (Description), students can now reuse a project description previously written by their group. I hope this change will make the individual work feel less overwhelming and more manageable.

Footnotes

1. The summary was generated by ChatGPT and then checked and edited by the examiner.