Hochschule Darmstadt - Fb Informatik

Drucken| Layout| Design| Schriftgröße English|
Module:Natural Language Systems
Module numbers:41.4268 [PVL 41.4269; Module 41.42680]
Study programme:JIM 2006 - Courses
Master 2006 - Katalog T: Theorieorientierte Module
Master 2004/2002 - Katalog WP: Anwendungskomponente
Master 2000 - Katalog B: Erweiterter Informatikbereich
Type of course:V+P = Lecture+Practical
Weekly hours:2+2
Credit Points:6
Exam:written exam
Registering for examexplicitly and independent of booking
PVL (e.g. Practical):graded (40% assignments + 30% practical course = 70% in total)
PVL percentage:70%
Frequency of offering:inactive
Required knowledge:it is nevertheless recommended to have some background in A.I.
Learning objectives:A primary goal of Artificial Intelligence is to enable computers to use natural language.
Applications of this capability include conversing with users to provide information or advice, translating from one language into another, comprehending, generating and summarizing text, and searching text for information relevant to some concern.
The approach taken in this course presumes that the ultimate success of any of these enterprises entails understanding and simulating a broad range of human cognitive capacities. Thus, while including more general issues of knowledge representation, meaning, common-sense reasoning (especially inference and planning) and knowledge organization and access the course emphasizes specifically linguistic concerns, such as grammar, parsing and generation. Other topics include lexical and grammatical disambiguation, the computational use non-literal language, and language acquisition.
The course covers a variety of approaches to these fundamental problems, but also examines practical techniques that implement partial solutions to problems such as lexical disambiguation and parsing, and some applications of these solutions to tasks such as information retrieval and machine translation.
While extensive familiarity with Artificial Intelligence or linguistics is not presumed, some background in AI is helpful.
Content:This course will cover the following aspects of Natural Language Processing (NLP):
In general, you will get an introduction to the following aspects by linking the "linguistics view" (computational linguistics) with the "artificial intelligence view" (natural language processing) at all times throughout the semester:
Morphology (= the analysis and generation of language on word level): e.g. problems with compounding and idiomatic phrases, homophonous strings as well as loan words and their processing using e.g. finite state automata as well as semantic networks. We will look at ambiguities in words like "pen" and "pipe", but will also discuss complex strings such as "Donaudampfschiffahrtskapitän" or language-mixes, such as "Er hat das Programm geupdatet."
Syntax (= the analysis and generation of language on phrasal and sentence level): e.g. applications such as machine translation and grammar checking and the processing using phase structure grammars as well as unification based formalisms, and relating those formalisms to recursive transition networks (RTNs) as well as augmented transition networks (ATNs).
Semantics (= language ambiguities on the level of "meaning"): represented by case structures and conceptual dependency structures. We will look at famous utterances such as: Colourless green ideas sleep furiously. And will discuss why the machine runs into problems during analysis, and how these problems can be overcome.
Speech Recognition: we will go over the algorithm used in speech recognition products (Voice Pro) and will discuss statistical language processing and it's theoretical background: Bayes Rules and Hidden Markov Models. In lab, we will get hands-on experience with speech recognition products.
Applicability:Several theroetical contents of the Bachelor studies find their applications in the field of natural language processing; such as finite state automata or basic algorithms in statistics (Hidden Markov Models and Bayes Rule).
By throroughly investigating such applications, the students will be able to draw the connection between theory and their value in applicative areas of Compuiter Science/Artificial Intelligence/Natural Language Systems.
Moreover, many of today's applications in the general field of Computer Science ask for natural language man-machine interfaces. The students will learn the various aspects that are involved to realize such interfaces by means of NLP technologies.
Literature:Jurafsky, Dan / Martin, James. Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics and Speech Recognition. Prentice Hall. 2000.
Manning, Christopher / Schütze, Hinrich. Foundations of Statistical Natural Language Processing. MIT Press. 1999.
Lecture style / Teaching aids:Lectures in seminar-style, i.e. discussions are welcome.
Script is available to students.
All lectures have been taped and are downloadable via the homepage of the instructor.
Responsibility:Bettina Harriehausen

[Fachbereich Informatik] [Hochschule Darmstadt]
© 2008 - 2022 FBI OBS Team. Alle Rechte vorbehalten.