Thomas Hanneforth, University of Potsdam
Robust speech and language technologies (shallow parsing, morphological analysis) are often based on finite state automata (FSA) because FSA have interesting mathematical properties which allow to build complex functionality in a modular way.
The topics of this course will be:
- Algebra and closure properties of finite state acceptors and transducers.
- Weighted FSA: using semiring structures to add weights to FSA
In the practical part of the course we try to build a robust parser based on finite state methods. Since we use the Potsdam Finite state machine library which is written in C++ a bit of knowledge in C++ would be helpful.
Course material