|
Course
description
Aims
The aim of this
module is to examine in detail some of the technologies of spoken
natural language processing, including grammars, parsing, spelling
correction, part-of-speech tagging, and spoken dialogue systems.
The practical component of the module involves the use of VoiceXML
to develop spoken dialogue systems.
Learning
Outcomes
Upon
the successful completion of this module a student should:
- be aware
of current research in spoken natural language processing
- understand
the main functional components of a typical spoken natural language
processing system
- have a detailed
knowledge of the models and algorithms used in natural language
technology, such as grammas and parsing
- understand
how these models and algorithms can be applied in spelling corrections
and part-of-speech tagging
- have practical
experience of the use of VoiceXML for spoken dialogue system development.
Course Topics
Introduction
to natural language processing
An overview
of natural language processing - application areas, current research
programmes, the relationship of natural language processing to other
disciplines, the components of a natural language processing system.
Spelling
correction
Minimum
edit distance.
Probabilistic methods for spelling correction.
Language
models
The use of n-gram language models in speech recognition and other
applications..
Part-of-speech
tagging
Rule-based and HMM based methods for part-of-speech tagging
Grammar and
Parsing
The basic concept of a linguistic grammar and its use in a spoken
dialogue system. Different parsing methods, including top-down,
bottom-up, and chart parsing.
Spoken Dialogue
Systems
Spoken dialogue systems are one of the main applications of speech
and language processing. In spoken dialogue systems all the components
of a speech and language processing system have to be integrated.
The main part of the course focusses on spoken dialogue systems,
examining issues such as: design methodologies, implementation,
dialogue control strategies. Practical work is based on VoiceXML.
Practicals
Practicals will take the form of laboratory based exercises to support
the course, using VoiceXML.
|
