Project seminar: Identifying Structure in Data

Module Description

Lecturer	Constantin Seibold
Module Type	Hauptseminar
Language	English
First Session	16.04.2026
Time and Place	Thursday, 13:15 - 14:45, INF 327 / SR 2
Commitment Period	tbd.

Participants

All advanced CL Bachelor students and all CL master students. Students from MSc Data and Computer Science or MSc Scientific Computing with Field of Application Computational Linguistics are welcome after getting permission from the lecturer. If the seminar should be oversubscribed, CL students will have priority.

Prerequisites for Participation

Basic knowledge of machine learning concepts and familiarity with Python are helpful but not required. Students will be guided through the necessary technical skills during the seminar. Students must be comfortable reading scientific papers in English. Prior coursework such as Einführung in die Computerlinguistik, Statistische Methoden für die Computerlinguistik, or Programmieren I/Fortgeschrittenes Programmieren für CL provides suitable preparation. If you are unsure about the preparation, contact the instructor.

Assessment

• Active participation in weekly supervisor meetings (20%)
• Final group presentation (~15 min + discussion, 30%)
• A jointly authored six-page conference-style research paper (50%)

Content

This project seminar covers computational methods for identifying and leveraging structure in data. Topics include clustering, dimensionality reduction, representation learning and infromation extraction, the use of language models for structure discovery, data visualization, knowledge discovery, novel class detection, and topic modeling.

Students work in groups of three on a semester-long research project with weekly supervisor meetings (~1-1.5h). Project topics will be presented in the first session for selection; students are also encouraged to propose their own project ideas within the seminar theme. Each group formulates a research question grounded in current literature, selects and applies appropriate methods to a real-world problem, and evaluates their approach experimentally. The seminar culminates in a final presentation and a six-page double-column paper.

Literature

General background and tutorials:

• Jurafsky & Martin, "Speech and Language Processing" (3rd ed., online): Link
• Goodfellow, Bengio & Courville, "Deep Learning" (MIT Press): Link
• scikit-learn User Guide: Clustering: Link
• scikit-learn User Guide: Manifold Learning & Dimensionality Reduction: Link
• McInnes et al., "Understanding UMAP" (interactive tutorial): Link
• BERTopic documentation and tutorials: Link

» More Materials