This project module will allow you to choose a research topic that has practical relevance and which falls in the broad area of human-centred AI, including natural language processing and information retrieval.
You will demonstrate that you can identify a suitable problem, devise and implement a solution and then evaluate it -- a bit like a mini Master dissertation.
One possibility is to extend and reposition the work you conducted last term in your group projects. We will also have more guest presenting ideas for projects. We will start in week 2 with an outline of interesting problems around knowledge extraction from social media presented by our partners at the City of Regensburg.
And (+++ hot off the press +++) there is a whole range of interesting challenges proposed for SemEval 2027.
Here are some other possible topics you might want to work on:
- Retrieval-augmented generation (RAG)
- Fake news detection
- Hate speech detection in social media
- Identification of biased content
- Dialogue systems, conversational search, chatbots
- Sentiment analysis
- Text summarisation
- State-of-the-art search engine techniques
An ideal outcome of this module could be paper that can be submitted to a conference.
Over the past few years this module resulted in several peer-reviewed publications and (as a motivation) we will start the module by looking at them!
I am looking forward to some interesting discussion of state-of-the-art human-centred AI problems with you. With any questions, please contact me directly via udo.kruschwitz@ur.de.
