Kurssikuvaus

Esitiedot:

• Experience of using Java with open source software (for example ctl257)
• Experience of writing XSLT transformations (for example clt232)
• Knowledge of general linguistics: morphology, syntax, semantics.

Tavoitteet:

• The course investigates the problems and techniques of natural language generation (NLG). Students learn about NLG tasks and architectures, both for text generation and for spoken dialogue response generation. Different approaches are compared, including template-based generation, deep vs. shallow generation, and XML-based generation.
• In a series of practical assignments, students will build simple natural language generation systems using Java, XML and open source software.

Kurssin sisältö:

• Introduction to Natural Language Generation.
• Canned-text generation.
• NLG tasks and architectures.
• Template-based generation.
• Deep vs. shallow generation.
• XML-based generation.
• Text planning.
• Microplanning: lexicalization.
• Microplanning: aggregation.
• Microplanning: referring expressions.
• Surface realization.
• Systemic Functional Grammar for NLG.
• Spoken dialogue systems.
• Examples of working NLG systems.

Kirjallisuus:

• Ehud Reiter & Robert Dale,
  Building Natural Language Generation Systems,
  Cambridge University Press, 2000.
  Contents and Chapter 1 (Introduction)

Kurssin laajuus opintopisteinä:

3 op

Opetus

Opettaja:

Graham Wilcock. The course is taught in English.

Suoritustapa:

The course consists of lectures, practical work, and a series of homework assignments. The course is passed by completing the assignments.

Paikka ja aika:

ma 11-14 S20A laitoksen unix-luokka, 15.1.-26.2.

Kurssimateriaali:

Kurssin verkkosivuilla, ks. sivupalkki.