Course web page, WS 2018 - 2019 SfS, University of Tübingen

This is the course page for the seminar course Sequence and Structure Learning in Computational Linguistics at the Department of Linguistics, University of Tübingen.

In a typical classification problem, the classes are simple labels with no internal structure. Some learning problems, however, require predicting structured outputs. This type of problems are prevalent in computational linguistics (CL), a typical example being predicting a parse tree for a given sentence. Sequences are another common type of structured output that are common in the CL literature.

In this class we will study methods of structured learning, focusing more on models of sequence learning, and their application in CL. This is a seminar course where we will read and discuss some classic, foundational papers, and some new, state-of-the-art papers on the topic.


This is an advanced seminar for Master’s and advanced Bachelor’s students. Prior exposure to machine learning and natural language processing is required. You can do with basic programming skills if you take the course for 6ECTS (see below), your should also be fluent in programming if you intend to take the course for 9ECTS.


The course is worth either 6 or 9 credits, depending on the workload. For 6ECTS, the evaluation will be based on active participation (Reading, and discussing the papers in the class and online. This may also include some small programming exercises), and leading the discussion on one of the papers. The 9ECTS option has the additional requirement of writing a term paper. The term paper is ideally based on a practical application of one of the methods discussed in the class.

We will use GitHub classroom for class discussion and scheduling. To be able to participate in the class discussion and access the material, you need to complete a trivial ‘assignment’. This step counts as registration to the course, please complete it by Wednesday 24, October 2018.