The course belongs to the MA Programme Linguistic Diversity in the Digital Age

• study track: language technology
• modules: Studies in Language Technology (LDA-T3100), Essentials in Language Technology (LDA-TA500), Comprehensive specialization in Language Technology (LDA-TB500)

This is an optional course.

The course is available to students from other study tracks and degree programmes.

• Programming for linguists or equivalent (BA level)
• Mathematics for linguists or equivalent (BA level)
• Machine learning for linguists or equivalent (BA level)

• Linguistics in the digital age
• Computational syntax
• Computational semantics
• Computational morphology

After successfully completing the course, students will be able to

• explain models and algorithms used in selected NLP applications
• describe properties of local prediction models and structural prediction models and methods that can be used to train them
• explain the differences between generative and discriminative models and between supervised and unsupervised learning
• describe the main components of a selected NLP application, for example a part-of-speech tagger
• train and evaluate practical NLP models in a sound and scientific manner.

Students are advised to take this course in year 2 (semester 3). The course is offered during the autumn term in period I.

Models and algorithms used in common NLP applications:

• Local prediction models and different training algorithms (Naive Bayes, perceptron, log-linear models)
• Structured prediction models and different training algorithms (HMM, structured perceptron, CRF)
• Dynamic programming algorithms for alignment and decoding
• Semi-supervised and unsupervised learning (EM algorithm).

Weekly lectures

Weekly assignments consisting of theoretical questions and programming exercises

The literature depends on the selected application, for example Philipp Koehn: "Statistical Machine Translation" (Cambridge University Press) in case of machine translation.

Other recommended literature: Manning and Schütze: Foundations of Statistical Natural Language Processing (MIT Press).

Additional web material and literature distributed on the course.

• Lectures and tutorials
• Interactive sessions, for example flipped classroom activities
• Problem-based collaborative project work
• Seminars with peer-review
• Activities documented in Moodle

Weekly assignments consisting of theoretical questions and programming exercises

• Contact teaching (lectures, tutorials, seminars)
• Self studies and group work

Examination:

• one or more of the following: flipped classroom activities, overview paper, written exam (part I)
• project report and presentation with peer-review (part II)