The 2026 episode at Faculty of Mathematics, Physics and Informatics of Comenius University
This site lists various ideas that could be explored as possible projects as part of the NLP course. The list is far from exhaustive, but represents (current) course instructor’s interests.
All of the ideas assume the final project will contain some “novelty bits”, either in the explored idea itself, its execution, practical usability or the underlying dataset.
Creating a new dataset for any of the NLP tasks listed below (or any other, really) is a huge plus.
Shared tasks#
Shared tasks are essentially “academic Kaggle”: you get a task, some data and produce a model that tries to do well on it. During the evaluation period, you normally produce a prediction on the test set. It’s a relatively straightforward way of going from a task to some solution, while not having to bother with the difficult part of finding an appropriate dataset. Further, there are highly likely other people working on the same task, so it’s really a bit of a “competition” (although that’s really not what it’s about and why it’s done).
CLEF 2026#
CLEF 2026 is the best fit for our semester: training data is available from February/March, with evaluation runs due May 7 and working notes due May 28. Registration is open until April 23. Here are some particularly relevant labs:
- CheckThat! Task 1: Source Retrieval for Scientific Web Claims Given a social media post referencing a scientific paper, retrieve the cited paper from a candidate pool (EN/DE/FR). (https://checkthat.gitlab.io/clef2026/task1/)
- CheckThat! Task 2b: Verdict Prediction Predict the factual status of numerical claims using LLM-generated reasoning traces. (https://checkthat.gitlab.io/clef2026/task2/)
- CheckThat! Task 3: Generating Full Fact-Checking Articles Given a claim, its veracity, and evidence documents, generate a complete fact-checking article with inline citations. (https://checkthat.gitlab.io/clef2026/task3/)
- EXIST: Sexism Identification in Memes Binary classification of whether a meme depicts sexism or criticizes sexist behavior (EN/ES). (https://nlp.uned.es/exist2026/#tasks)
- ELOQUENT: Voight-Kampff Determine whether text is machine-generated or human-authored – can your LLM fool a classifier? (https://eloquent-lab.github.io/task-voight-kampff/)
- ELOQUENT: Cultural Robustness and Diversity Examine whether an LLM responds consistently across languages and cultural contexts – run a model in 5+ European languages and submit outputs. (https://eloquent-lab.github.io/task-cultural-robustness/)
- ELOQUENT: Topical PISA Quiz Generate exam questions from a text, answer them, or score student responses – flexible participation, pick any one of the three subtasks. (https://eloquent-lab.github.io/task-pisa-generate-items/)
- SimpleText: Sentence-level Scientific Text Simplification Simplify individual sentences from scientific papers. (https://simpletext-madics.github.io/2026/tasks)
- TalentCLEF Task A: Contextualized Job-Person Matching Rank candidate resumes by relevance for a given job offer (EN/ES/cross-lingual). Runs due May 3. (https://talentclef.github.io/talentclef/docs/talentclef-2026/task-summary/)
- eRisk: Early Detection of Depression Analyze sequentially arriving conversations and flag depression risk as early as possible – a streaming/online classification setup. Runs due April 22. (https://erisk.irlab.org/)
- eRisk: ADHD Symptom Sentence Ranking Rank sentences by relevance to the 18 ADHD symptoms from the ASRS-v1.1 scale – a zero-shot retrieval task with no training data. Runs due April 15. (https://erisk.irlab.org/)
- HIPE: Person-Place Relation Extraction Extract who was where from noisy OCR-derived historical texts – designed to work well with LLMs. (https://hipe-eval.github.io/HIPE-2026/)
- JOKER: Pun Translation Translate puns and jokes across languages while preserving comedic intent and wordplay structure. (https://www.joker-project.com/)
Or any other CLEF 2026 lab!
NTCIR-19#
NTCIR-19 (Tokyo, December 8–10) has formal run submissions in June–July 2026.
- R2C2: Answering with Confidence Provide an answer with a calibrated confidence score – say “I don’t know” when confidence is low rather than returning an unreliable answer. Submission: June 1. (https://research.nii.ac.jp/ntcir/ntcir-19/tasks.html)
- Tip-of-the-Tongue Known-item retrieval: find something when you can only vaguely describe it, not name it. Submission: June 1. (https://research.nii.ac.jp/ntcir/ntcir-19/tasks.html)
- FEHU: Human Value Recognition Multi-label classification of human values (equality, justice, freedom of thought, etc.) expressed in news articles. Submission: June 1. (https://research.nii.ac.jp/ntcir/ntcir-19/tasks.html)
IberLEF 2026#
IberLEF 2026 runs shared tasks in Spanish and other Iberian languages (workshop: September 22). A good fit if you have some Spanish.
- HAHA: Humor Detection Determine whether Spanish news headlines are satirical jokes or real news. System deadline: June 3. (https://www.fing.edu.uy/inco/grupos/pln/haha/)
- HAHA: LLM-generated Humor Detection Identify whether jokes inspired by news headlines were created by an AI or a human. System deadline: June 3. (https://www.fing.edu.uy/inco/grupos/pln/haha/)
Or any other IberLEF 2026 task!
Project MIMEDIS#
The project’s aim is to “study the impact of media discourse on attitudes towards migration, migrants and migration policy in Slovakia”. As such, there are many classification tasks that can be explored in that regard.
You can find more about the project at https://cogsci.fmph.uniba.sk/MIMEDIS/index.html.
Your own idea!#
Feel free to come up with an idea on your own -- if you are working on something NLP-related for your thesis, that would be a good candidate. But in general, I’d be happy to talk about any NLP-related idea you may have.
Alternatively feel free to check out the sites below, find an NLP task you find interesting and see if you can make an interesting project out of it!