On the Resolution of Bridging References within Information Extraction Systems

contributor	Please contact
creator	Cimiano Lavin, Philipp
date	2002

description	163 pages
	This diploma thesis presents an ontology driven approach to bridging reference resolution within information extraction systems. In contrast to general NLP tools, information extraction systems are typically restricted to a specific domain so that a model of the domain in form of an ontology can be created and used within bridging reference resolution. The thesis presents a formal definition of what an ontology is as well as a shallow reasoning procedure on the basis of this definition. Furthermore, it explores how an ontology as declarative representation formalism of domain knowledge as well as the above mentioned reasoning mechanism can be exploited within bridging reference resolution. The application domain of the developed bridging reference resolution approach are short texts of a biochemical nature and in particular SWISS-PROT database entries describing the function of a certain protein. The principles under which an ontology for the domain in question can be developed are discussed and the approach is evaluated against a training and test set of the above mentioned protein function descriptions.
format	application/pdf
	730689 Bytes

identifier

language	eng
publisher	Stuttgart, Germany, Universität Stuttgart
relation	Diploma Thesis No. 2025
source	ftp://ftp.informatik.uni-stuttgart.de/pub/library/medoc.ustuttgart_fi/DIP-2025/DIP-2025.pdf
subject	Knowledge Representation Formalisms and Methods (CR I.2.4)
	Natural Language Processing (CR I.2.7)
	Life and Medical Sciences (CR J.3)
	bridging references
	discourse analysis
	information extraction
	applications of ontologies
	ontological reasoning
	knowledge representation
title	On the Resolution of Bridging References within Information Extraction Systems
type	Text
	Diploma Thesis