Project
This page extends the course memo with more information specific to the project work in this course.
Overview
In the project module, you work individually on a substantial text mining problem of your choice. Identifying and formulating a topic is part of your work, as is implementing suitable methods and analysing your results. The project is examined by a written project report that you hand in for one of the course’s examination dates.
An important idea behind the project module is that you can get help from the examiner. You can get help by contacting the examiner via e-mail or by booking a meeting. To give every student the chance to discuss their project, these meetings are strictly limited to 15 minutes each. You can use this time however you like, but please come prepared with concrete questions!
- This option is only available during the course period, starting in week 46, with some break around the holidays.
- From week 51 onwards, I will only offer remote meetings (via Zoom).
- It is okay to book more than one meeting throughout the course, but please do not book more than one meeting at the same time or in the same week.
Finding a topic
You are free to work on any topic within the area of text mining. The lectures and labs should offer some inspiration on possible topics to work on, but we also provide a selection of example projects from previous course iterations. Note that it’s perfectly fine to work on the same topic as a project from previous years!
Finding resources
Whatever topic you choose, you will probably need to decide on which data to use, what methodology to apply, and how to evaluate your results. Here are some pointers that may be useful along the way; we will also discuss them during the project kick-off session. As mentioned before, you are encouraged to book a meeting with the examiner to discuss any questions, ideas, or problems you may have.
Places to find data
- Kaggle
- HuggingFace Datasets
- Riksdagens öppna data (in Swedish)
Places to find research papers
- ACL Anthology, hosts most of the biggest NLP conferences and journals
- Semantic Scholar, a more general research tool for scientific literature
- Scopus AI, licensed through LiU’s library
- Elicit (requires registration), an AI-based research assistant that will link to relevant academic papers
Further resources
If your project uses machine learning in any way, we recommend the following paper to help you avoid some common issues:
- How to avoid machine learning pitfalls: a guide for academic researchers, by Michael A. Lones.
Accessing GPUs
Many more advanced methods for text mining require access to a GPU. You can use the GPU-equipped lab rooms in B-huset – Åsgard, Egypten, Olympen – for this purpose, preferably during the times we have specifically booked them for this course. You can find these in the course schedule under “individual project work”; these are reservations for you to work individually on your project, with no teachers present. Alternatively, you can also try services like Colab or Kaggle to run your experiments.