direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Inhalt des Dokuments

KIT-Reports

KIT Report 160

Vertex Disjoint Path in Upward Planar Graphs
Saeed Akhoondian Amiri, Ali Golshani, Stephan Kreutzer, Sebastian Siebertz
Dezember 2013
To download the pdf version of this report click below: 
Download PDF Version. (PDF, 463,7 KB)

KIT Report 159

Directed Elimination Games
Viktor Engelmann
Dezember 2013
To download the pdf version of this report click below: 
Download PDF Version. (PDF, 429,1 KB)

KIT Report 158



Epsilon-mu-logics as a theory of propositions
Sebastian Bab
January 2012, 26 pages
To download the pdf version of this report click below:
Download PDF Version. (PDF, 277,9 KB)

KIT Report 157


Abstraktion in der Abbildungstheorie

Andrea Hillenbrand

August 2009, 47 pages
To download the pdf version of this report click below:
Download PDF Version. (PDF, 283,5 KB)

KIT Report 156


Modelle als Akteure. Fallstudien

Bernd Mahr, Reinhard Wendler

February 2009, 79 pages
To download the pdf version of this report click below:
Download PDF Version. (PDF, 4,0 MB)

KIT Report 155


ε-semantische Modellierung eines Modells der Auffassung
Florian Eilers

February 2009, 40 pages
To download the pdf version of this report click below:
Download PDF Version.

KIT Report 154

Ein Modell der Open-Source-Entwicklung
Steffen Evers
Mai 2008, 163 pages
To download the pdf version of this report click below:
Download PDF Version.

KIT Report 153

Zur Frage korrekter und vollständiger Kalküle der Epsilon_mu-Logiken
Sebastian Bab
September 2007, 36 pages
To download the ps version of this report click below:
Download PS Version.
To download the pdf version of this report click below:
Download PDF Version.

KIT Report 152

Stil und Modell
Bernd Mahr
April 2006, 53 pages
Über den Zusammenhang von Stil und Modell in der Bildenden Kunst und der Programmierung
To download the pdf version of this report click below:
Download PDF Version.

KIT Report 150

Wissen im Modell

Bernd Mahr
December 2004, 21 pages
To download the pdf version of this report click below:
Download PDF Version.

KIT Report 149

Computer-Aided Representation, Analysis and Conveyance of Musical and Music-Theoretical Structures

Thomas Noll, Monika Brand, Jörg Garbers, Andreas Nestke, Anja Volk
March 2004, 52 pages
Final Report of the Interdisciplinary Research Group KIT-MaMuTh for Mathematical Music Theory at the Technical University Berlin (1998-2003). This report is composed as a >guided tour< trough the various results of our five-year research work, which are available in form of publications, manuscripts and software. Through this guidance the reader may better understand how these results are connected with one another and how they emerged. After a short portrait (section 1) of the group and its members each of the subsequent sections is devoted to one of our research directions, obtained results and underlying questions. Section 2 provides an introduction to Mathematical Music Theory and motivates the particular choice of research subjects and discusses their mutual cross-relations on the conceptual level. Section 3 traces our activities on the level of software development. The notion of (musical and music-theoretical) structure around which our threefold research subject "Computer-Aided Repesentation, Analysis and Conveyance of musical and music-theoretical Structures" is formed motivates the application of structural mathematics in various ways. Each of the three sections 4, 5 and 6 is dedicated to the structures of our investigation on a specific music-theoretical level of description: metric structure, melodic structure and harmonic structure. Somewhat hidden behind this emphasis of structure - but equally central to our work - is the notion of transformation. Simply speaking, to understand a structure requires its transformation. While applying this principle to the investigation of metric, melodic and harmonic structure we implicitly - and later explicitly - were concerned with issues of processuality as well. This is documented in section 7.
To download the pdf version of this report click below:
Download PDF Version.

KIT Report 148 <in German>

Kohärenzrelationen: Ein Vergleich von Kontrast und Konzession

Carla Umbach, Manfred Stede
1999, 17 pages
This paper is concerned with the coherence relations of Contrast and of Concession which are employed, e.g., in Rhetorical Structure Theorie. It will be discussed (a) whether there are different kinds of Contrast relations, and (b) whether the Concession relation is also a special case of Contrast. An analysis of Contrast will be proposed which is based on focus semantics. This analysis provides a unified account of Contrast subsuming semantic opposition and also denial of expectation. The relation of Concession, however, will be shown to be distinct (though compatible). Finally, the status of discourse relations and their role in Natural Language Generation are discussed.
To download the pdf version of this report click below:
Download PDF Version.

KIT Report 147 <in German>

KIT-MaMuTh: Zwischenbericht Softwareentwicklung

Jörg Garbers
December 2001, 26 pages
This report presents software activities that took place in the interdisciplinary research projekt ''Interdisciplinary Research Group KIT-MaMuTh for Mathematical Music Theory'' by the Technical University Berlin (1998-2001).
 
The report comprises two chapters. The first one describes activities within the software packages Rubato, OpenMusic, Humdrum and JavaView, with emphasis on the portation of Rubato from NeXTStep to the Mac OS X operating system. The second one describes activities that enable the pairwise functional and user interface integration of these software packages.
To download the pdf version of this report click below:
Download PDF Version.

KIT Report 146 <in German>

KIT-MaMuTh: Zwischenbericht

Thomas Noll
December 2001, 41 pages
To download the pdf version of this report click below:
Download PDF Version.

KIT-Report 144

dSelf - A distributed SELF

Robert Tolksdorf, Kai Knubben
2001, 12 pages
dSelf is an extension to the delegation- and prototype-based object-oriented language SELF. It adds distributed objects and transparent remote reference resolution to the languages. As a consequence, dSelf facilitates distributed inheritance and instantiation mechanisms. We describe the current conception and implementation of dSelf.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 143< in German >

"Kontextauswertung" im Verbundprojekt Verbmobil II:
Multilinguale robuste und direkte Übersetzung spontansprachlicher Dialoge

Bernd Mahr, Stephan Koch, Uwe Küssner, Manfred Stede, Dan Tidhar
2001, 33 pages
To download the ps version of this report click below:
Download PS Version.

KIT-Report 141< in German >

Prinzipien der Kontextualisierung

Bernd Mahr, Klaus Eyferth, Roland Posner, Fritz Wysotzki
1997, 122 pages
This report presents findings that resulted from the interdisciplinary research projekt ''Kognition und Kontextfunded by the Technical University Berlin (1993-1996).
 
The volume comprises five individual contributions on the issues of context-dependency and perspectivity discussing these topics from the point of view of cognitive psychology, semantics, artificial intelligence and theoretical computer science.
 
In addition, each paper is followed by comments made by the remaining authors pointing out interdisciplinary relations. Finally, there is a reply to these comments by the author himself.
To download the pdf version of this report click below:
HTML Version

KIT-Report 140< in German >

Termpräzisierung - kontextuelle Steuerung der Interpretation durch Apposition und Typisierung

Carla Umbach
1996, 191 pages
This report focusses on the problem of context-dependency in natural language interpretation. Based on the theory of precisification a semantic model is developed that accounts for both declarative and procedural aspects of the interpretation of a lexical item. As a linguistic guideline the semantics of appositions are investigated. Formal semantics are given in a first order framework with a dynamic notion of type assignment.
 
It will be concluded that (1) In modelling context-dependency indefiniteness because of too little contextual information and indefiniteness because of too much (i.e. inconsistent) contextual information have to be handled separately. In the model proposed here the former is represented by quantifying over possible refinements of interpretation, whereas the latter results in a type clash. (2) The main semantic feature of appositions consists in triggering a presupposition supplying contextual information. The meaning of an apposition can formally be represented as a (dynamic) type assignment. (3) There is a mutually informative connection between the notion of a presupposition and the mechanism of typing both providing contextual information to control interpretation.
To download the pdf version of this report click below:
HTML Version.

KIT-Report 139

Semantische Auswertung für die Übersetzung - Abschlußbericht Projekt KIT-VM11

J.Joachim Quantz, Bernd Mahr, Birte Schmitz, Uwe Küssner, Manfred Stede, Guido Dunker, Frank Bergmann, Ivonne Kellner
June 1996, 30 pages
This report is the final report of the project KIT-VM11. KIT-VM11 was part of the first phase (1993-1996) of the VERBMOBIL project, a project concerned with face-to-face dialogue interpreting funded by the German Ministry of Research and Technology. The paper describes the task of semantic evaluation in the VERBMOBIL system, i.e. the processing of the Verbmobil Interface Term (VIT), the domain model and disambiguation, dialogue act recognition, temporal reasoning, the description logic system FLEX++ and the architecture of the semantic evaluation modul in the VERBMOBIL system.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 138

Classifying Verbs of Perception. An Exercise in Classification-Based Phrase Structure Grammar (Beiträge zur klassifikationsbasierten Phrasenstrukturgrammatik III)

Michael Grabski
December 1996, 77 pages
A Contribution to a German fragment of `Classification-based Phrase Structure Grammar' (CPSG), developed in the research project `Kognition und Kontext' (Technical University of Berlin), is documented in this volume. CPSG is an approach similar to HPSG, that has been proposed by Richard Cooper (1990); it contains a more fully developed semantics than standard HPSG and specifies a separate `empirical domain', which allows transfer of standard semantic descriptions into feature structure representations.
 
Syntactic and semantic properties of expressions are represented in CPSG in terms of type hierarchies. A semantic classification for a set of perception verbs is discussed in this paper and represented in feature structures. Aspects considered here are lexical decomposition, contextual informativity and `field properties' of the lexical items.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 137< in German >

Beiträge zur klassifikationsbasierten Phrasenstrukturgrammatik II:
CPSG im Vergleich zu ALE und CUF

Stefan Havenstein, Chris Werner
December 1996, 96 pages
This reports includes two papers comparing CPSG (Classification-based Phrase Structure Grammar) with ALE (Attribute Logic Engine) and CUF (Categorial Unification Formalism), respectively. CPSG is an approach similar to HPSG, that has been proposed by Richard Cooper (1990). In the project ``Kognition und Kontext'' a German CPSG-fragment has been developed focussing on semantic properties of verbs of perception. Both papers describe a reimplementation of this CPSG-fragment in ALE rsp. CUF and discuss the pros and cons for natural language proccessing.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 136< in German >

Die Verben des Riechens im Deutschen und Englischen: Eine kontrastive semantische Analyse

Ellen Fricke
November 1996, 144 pages
The aim of this work is to give equal weight both to reflections on the theoretical basis of lexical field theory and to the descriptive comparative analysis of olfactory verbs in German and English within this framework.The lexical field theory is guided by he principle that the meaning of a lexeme is determined by the network of paradigmatic and syntagmatic relations that hold between the lexeme in question and its neighbours in the same lexical field. Two approaches of lexical field theory, Eugenio Coseriu's on the one hand and Rolf Peter Lutzeier' on the other, are discussed, especially with regard to their problems of application to empirical data. A modified scheme of analysis based on this comparison is proposed which takes the role of the verbal context into consideration.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 135< in German >

Kognitionswissenschaft an der Technischen Universität Berlin:
Ringvorlesung im Wintersemester 1995/1996

(eds.):
Bernd Mahr, Klaus Eyferth, Roland Posner, Fritz Wysotzki, Ute Schmid

September 1996, 21 pages
The lecture 'Cognitive Science at the Berlin University of Technology' has been established in the winter term of 1995/1996. It was introduced as a forum where scientists of different fields of research can present and discuss their work in the context of cognitive science. This report presents the topics covered in the first cycle of the lecture.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 134< in German >

Der Dienstbegriff in den ISO-Referenzmodellen OSI und ODP

Thorsten Albrecht
December 1996, 65 pages
The interaction between components of distributed information processing systems is often referred to as an offer or an use of a service. The different viewpoints by which systems and their components can be described and modelled influence this concept of service. In this case the concept of a service is usually not to be seen as a complex service in an economical sense. Different aspects of a service, in the context of general system modelling of distributed systems, are provided by two ISO standards, the Reference Model of Open System Interconnection (RM-OSI) and the Reference Model of Open Distributed Processing (RM-ODP). In this report, the service aspects of both models are analysed. After an introduction to the general modelling concepts of both standards, the respective service specific concepts are presented. The service aspect of RM-OSI is referred to a communication service, which as basic infrastructure allows and influences the interactions between system components. The components interactions cannot be modelled in terms of RM-OSI as an independent and complex service and is out of scope of the model. In contrast to that, RM-ODP makes possible to describe any complex system components and even whole systems as objects. Interactions of objects can be modelled in terms of behavioural concepts together with contractual concepts. A consequence of this modelling technique is the possibility to express basic communication services in terms of RM-OSI as well as approaches to more complex economic influenced services.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 133

Using the BACK System in MIHMA

A. Schmiedel, P. Volle, T. Hoppe, C. Kindermann, H. Oertel, O.K. Paulus
April 1996, 14 pages
The goal of the MIHMA project was to develop a method and a technology for building integrative information services which rely on semantic integration of references to on-line information into a domain model. This approach was demonstrated by an operational information service prototype for the domain of leukemia and bone marrow transplantation.
 
This domain model is the central component supporting semantic integration. It enhances the set of relevant sources with an additional unifying layer of domain knowledge, supporting ``semantic navigation'', i.e., navigation based on a model of information sources and the subjects they cover. This model contains a representation of the domain of the information service (in the case of the BMT Line application, medical topics related to bone marrow transplantation and relationships between them), as well as information sources themselves and other relevant entities such as persons or organizations, and of user profiles, i.e., sets of preferences defined by users subscribed to the service.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 132

The MIHMA Demonstrator Application: bmtline

C. Kindermann, T. Hoppe, D. Marwinski, H. Oertel, O.K. Paulus, E. Buu, S. Heimann, A. Schmiedel, R. Tolksdorf, P. Volle
April 1996, 37 pages
Despite the general availability of information on the Internet, users find it difficult to locate a particular piece of information. To locate information requires users to visit many different servers, and to browse their respective contents, which is time consuming, confusing, and inefficient. To locate useful information in the Internet's information space by navigation only is so laborious that it severely limits the Internet's usefulness as a source of information.
 
In this report, the development of ''bmtline'', a Web-based medical information service is described that directly addresses the scenario presented above. ''bmtlineprovides a reference database for Web and Internet resources in the domain of bone marrow transplant and leukemia. The added value of ''bmtlineïs based on the provider controlled content of the information it references, the high degree of semantic integration of referenced material which is based on a complex semantic model, the support of different standard user classes, and the support of individual customization of the service through the application of single users' profiles.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 131

The MIHMA Project Technology Report

P. Volle, A. Schmiedel, E. Buu, S. Heimann, T. Hoppe, D. Lutzebäck, O.K. Paulus, C. Kindermann, H. Oertel, R. Tolksdorf
April 1996, 28 pages
The increasing amount of data available in the World Wide Web has induced the construction of information services whose primary purpose is not to provide original content, but rather to ease the access to information. However, they do not always provide satisfactory answers to a user's need, especially when one is interested in a particular topic relevant to a complex domain, such as a technical or scientific domain. What is generally expected in this case is a selection of relevant information, organized in a way that reflects knowledge of the domain, i.e., that allows retrieval using relevant conceptual categories and terminology in use in the domain.
 
The goal of the MIHMA project was to contribute to a solution to this problem. We addressed it by developing a method and a technology for building integrative information services which rely on semantic integration of references to on-line information into a domain model. This approach was demonstrated by an operational information service prototype for the domain of leukemia and bone marrow transplantation.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 130

The MIHMA Project - Overview Report

O.K. Paulus, P. Volle, E. Buu, S. Heimann, T. Hoppe, C. Kindermann, D. Lutzebäck, D. Marwinski, H. Oertel, A. Schmiedel, R. Tolksdorf
April 1996, 23 pages
The aim of the MIHMA project was to develop mehtods and tools to build integrative on-line information services. This goal has been achieved and demonstrated by an operational prototype system for the domain of leudemia and bone marrow transplantation (BMT). The MIHMA approach is based on semantic integration of references to on-line information into a complex domain model. This process is supported by several automated tools for knowledge base management, document acquisition, hypertext generation and reference maintenance. Individual customization is supported through the definition of user profiles.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 129< in German >

Beiträge zur klassifikationsbasierten Phrasenstrukturgrammatik I.

Michael Grabski, Ellen Fricke, Stefan Havenstein, Heike Pisch
July 1996, 141 pages
Ongoing work on a German fragment of a "Classification-based Phrase Structure Grammar" (CPSG) that is being developed in the research project "Kognition und Kontext" (KuK) is documented in this volume. CPSG is an approach similar to HPSG, that has been designed by Richard Cooper (in his Edinburgh PhD Thesis); it contains a more fully developed semantics and specifies a separate empirical domain, thereby addressing transfer of existing grammatical descriptions into feature structure representations. The fragment developed in KuK has been centered around the semantics of verbs of perception, as the affinity of these expressions to notions of context and perspective has been of specific interest to the project. Syntactic and semantic properties of expressions are represented in CPSG in terms of type hierarchies. Subcategorisation properties of verbs are dealt with, in this way, in the contribution by E.Fricke and H.Pisch. A number of semantic classifications, for a set of perception verbs, is proposed in the contribution by M.Grabski. A documentation of the fragments implementation is given by S.Havenstein and H.Pisch.(A second volume is planned.).
To download the ps version of this report click below:
Download PS Version.

KIT-Report 128< in German >

Der Computer - Konstruktion von Realität durch Schrift?

Martin Fischer
July 1996, 100 pages
This paper deals with questions about computers as a medium and the connection between computers and writing.
 
In it a concept of writing is developed based on Nelson Goodman's symbol theory. The structure of symbol systems is used as the criterion for the classification of these systems. In opposition to the thesis of new orality, it is argued that computers and writing are inseparable.
 
Starting with the views of computer scientists on semantics, it is shown that virtual realities and other computer-based artifacts are reality constructions based on formal written models.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 127< in German >

Sprachliche Perspektiven und ihre Repräsentation.
Einige semantische Szenarios

Michael Grabski
May 1996, 40 pages
Perspective can be seen as determined by two interaction sorts of information according to a theory of Barwise: 1. descriptive information as encoded in relations and 2. contextual information that determines what part of the world is focussed by some agent of an inquiry. Descriptive information makes up a perspective if it is confined to some such focus situation; i.e. it does not hold in larger situations, that embed the focus situation. That is, it is non-persistent. General conditions for this configuration of situated facts are stated and applied to some differing semantic phenomena in natural language, among them the semantics of indexicals, of quotations, and of a type of thematic roles. For the latter case, ideas from a theory of event perception by Warren and Shaw are employed. A situation theoretic frame is used for modelling.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 126

Object Access in Mental Models under Different Perspectives
Induced by Linguistic Expressions

Robin Hoernig, Berry Claus, Daniel Durstewitz, Ellen Fricke, Ute Schmid, Klaus Eyferth
April 1996, 22 pages
This report replaces our report no. 121.
A mental model of a text should allow for access to objects distributed spatially, according to the text. An experiment was conducted to test three hypotheses about the representation of spatial relations between a recipient oftexts and the objects mentioned. Subjects read stories about a protagonist situated in a room, describing several objects in specified spatial relations to the protagonist. After each story the subjects decided on two objects whether these were mentioned in the text (target) or not (distractor).
 
First hypothesis: The latency of target verification depends on the spatial relation between the protagonist and the object. Objects in front of the protagonist should be verified faster than those beside and behind him. If objects beside or behind the protagonist are verified faster in a recognition task, it should allow one to decide between one of two alternative models: the spatial framework model (objects beside are slowest) and the model of mental transformation (objects behind are slowest). The results corresponded to the latter. This seems to suggest that the access to mentioned objects depends on amental model of the situation which is transformed analogous to a perceptual representation.
 
Second hypothesis: If the protagonist changes his orientation, according to thetext, the previous relations will be rotated. After the answer to the first test object the story was continued and a new orientation of the protagonist was mentioned. The verification latency of the second test object was determined definitely by the new orientation. This result supports an access toobjects via a mental model.
 
Third hypothesis: 'Experiencer' verbs (e.g., ''Peter sees the wardrobe.'') are more likely to induce the perspective of the protagonist than 'agent' verbs (e.g., ''Peter looks at the wardrobe.''). The latter version should support an external perspective of the recipient. A systematic structure in latencies of object verification will be found only under the protagonist's perspective. In case of an external perspective, response times should be about equal for all positions of objects. The response times confirmed these hypotheses. Finally, the categories 'experiencer verbs' and 'agent verbs' proved to induce differentperspectives, according to the hypothesis.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 125< in German >

Abbild oder Konstruktion
- Modellierungsperspektiven in der Informatik -

Martin Fischer, Gernot Grube, Fanny-Michaela Reisin
December 1995, ca. 140 pages
This report contains some results of a student project on modeling. We introduce the representation view and the construction view as two complementary views of modeling in computer science. By interpreting basic philosophical and social science texts according to constructivistic approaches and approaches based on the so-called correspondence theory of truth, the two views are motivated, explained, and related to some questions in computer science. Then problems, modeling approaches, and modeling methods in the field of software engineering and artificial intelligence are analyzed, interpreted and classified in terms of the complementary views of representation and construction.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 124

The FLEX System

J. Joachim Quantz, Guido Dunker, Frank W. Bergmann, Ivonne Kellner
December 1995, ca. 150 pages
This report describes the Description Logic (DL) system FLEX. It consists of a brief overview over the field of Description Logics in general and the characteristics of FLEX, a tutorial for the FLEX system, a brief description of the inference algorithms, and an appendix containing a syntax overview, the formal semantics, a reference manual, and an installation guide.
 
In a sense, the FLEX system can be seen as an extension of the DL system BACK. The main differences are that FLEX supports full disjunction and negation, weighted defaults, situated object descriptions, term- valued features, and flexible inference strategies. On the other hand FLEX does not support some of the functionality provied by the BACK system, such as revision, for example.
 
The FLEX system is developed in the project KIT-VM11, which is part of the VERBMOBIL project, a project concerned with face-to-face dialogue interpreting funded by the German Ministry of Education, Science, Research and Technology. Our main criteria for designing the FLEX system are therefore based on requirements arising from the application of semantic disambiguation in Machine Translation.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 123

A Framework for Telemedicine Services in Europe - An Essential Guide for Establishing Telemedicine Services

Marlene Gerneth, Horst Hansen, Bernd Mahr, Christine Seidel, Gavin Venters, Howard Williams (eds.)
March 1995, ca. 1.000 pages
The FEST Guide is a document guiding to the Framework for Telemedicine Services developed by the FEST project. It gives guidance to those who are interested in telemedicince and allows the benefits of applying the Framework to be realised by a wider audience. As a usage oriented guide, the intention of this document is to present the FEST Framework in an easily digestible form. It does so by packaging the vast variety of results of FEST in a comprehensive way so that they are better understandable and accessible. To this end, the main part of the document focusses on the usage of the Framework. Supplementing this, nine appendices A to I contain the technical and theoretical details of the Framework (e.g. the Question Set and the Body of Information).
To download the pdf version of this report click below:
Ordner

KIT-Report 122

The VERBMOBIL Domain Model

J. Joachim Quantz, Manfred Gehrke, Uwe Küssner, Birte Schmitz
December 1994, 69 pages
This report describes the domain model used in the German Dialogue Interpreting project VERBMOBIL. In order to make the design principles underlying the modeling explicit, we begin with a brief sketch of the VERBMOBIL demonstrator architecture from the perspective of the domain model. We then present some rather general considerations on the nature of domain modeling and its relationship to semantics.
 
We claim that the semantic information contained in the model mainly serves two tasks. For one thing, it provides the basis for a conceptual transfer from German to English; on the other hand, it provides information needed for disambiguation. We argue that these tasks pose different requirements, and that domain modeling in general is highly task-dependent. A brief survey of domain models or ontologies used in existing NLP systems confirms this position.
 
We finally describe the different parts of the domain model, explain our design decisions, and present examples of how the information contained in the model can be actually used in the VERBMOBIL demonstrator. In doing so, we also point out the main functionality of FLEX, the Description Logic system used for the modeling.
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 120 < Mostly in German >

Ambiguity and Strategies of Disambiguation

J. Joachim Quantz, Birte Schmitz (eds.)
December 1994, 166 pages
On the occasion of the annual meeting of the Deutsche Gesellschaft für Sprachwissenschaft in March 94 we organized a workshop on ambiguity and disambiguation strategies. This report contains seven of the contributions presented at the workshop (mostly written in German). In organizing the workshop we aimed at bringing together researchers from such different fields as computational linguistics, artificial intelligence, machine translation, theoretical linguistics, psychology, and translation theory.
 
Birgit Apfelbaum uses a corpus of German-French interpreting situations as a basis for investigating human strategies towards disambiguation. Michael Grabski discusses a specific aspect of polysemy in the context of translation, namely cases where a lexical distinction in the source language is not made in the target language. Klaus von Heusinger proposes a formal framework in which definite and anaphoric noun phrases as well as anaphoric pronouns can be uniformly represented. Bernhard Kipper discusses ambiguities arising in the interpretation of modal verbs. Lars Konieczny, Barbara Hemforth, and Nicole Voelker investigate human strategies to resolve ambiguities stemming from PP-attachment. Birte Prahl investigates human translation strategies in a situation where only limited contextual information is available. Manfred Stede describes an ambiguity problem, namely the lexicalization of the ``substitution'' relation (e.g. `but', `instead', `rather') from the perspective of language generation.
 
In our introduction we briefly sketch our own perspective on the problem of ambiguity, distinguish various aspects of ambiguity, indicate which aspects are addressed by the articles, and finally summarize their contents.
To download the pdf version of this report click below:
Ordner

KIT-Report 119 < in German >


Die Verbindung von mikro- und makrostruktureller Information zur automatischen Erkennung von Sprechhandlungstypen



Birte Schmitz

 
December 1994, 13 pages
 
 
The results described here have been developed within the field of automatic interpreting. We focus on the analysis of spoken language with dialogue processing methods.
 
We demonstrate that for an adequate translation of most of the verbs occurring in appointment scheduling dialogues the speech-event types of their utterance is relevant. We propose an approach for the automatic assignment of speech-event types to utterances. This approach applies weighted preference rules that are based on micro as well as macro-structural information.
 

KIT-Report 118 < in German >


Termpräzisierung als Mittel der lexikalischen Desambiguierung in der Sprachverarbeitung



Carla Umbach

 
November 1994, 24 pages
 
 
An ambigous lexical item is said to have different readings with respect to its meaning. This is slightly misleading because it suggests that there are a distinct number of readings. However, beyond the limited number of conventional readings one can always imagine others and, moreover, a reading itself may be split up into different variants depending on the semantic granularity intended. Lexical disambiguation then can be seen as a process that stepwise reduces semantic indeterminacy thus making the sense of the lexical item more precise in depth as well as in breadth. This process usually has to rely on the information implicitly given by the context.
 
In this paper, it is shown that German apposition with als and wie can be regarded as linguistic means to make the sense of a lexical item more precise according to its intended meaning. It is proposed then to exploit the semantic characteristics of als- and wie-apposition as a guideline for ``precisation-like`` operations in the NLP disambiguation of lexical items. Such operations will facilitate variable depth of analysis because they allow for the filling in of contextual information according to the degree of preciseness required by the respective processing task. Formal characteristics of these operations are discussed within a type-theoretically motivated framework.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 117


A Proof Theory for Preferential Default Description Logics



Sven Suska

 
July 1994, 68 pages
 
 
An extension of Description Logics by weighted default rules is presented, using the semantics of preferential models. Such Preferential Default Description Logics adhere to the principle of exception minimization and are therefore suited for certain tasks in Natural Language Processing.
 
A sound and complete proof theory is given, based on establishing a correspondence between models and sets of default applications (`default spaces'). A Prolog algorithm computing maximal default spaces is described and formally verified.
 
As a consequence, the default extension is decidable if the underlying Description Logic is. However, its complexity is exponential, some ideas are proposed to cope with this problem. A tractable subclass is characterized, for which a polynomial algorithm is sketched.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 116 < in German >


Installations- und Benutzungs- handbuch des experimentellen MÜ-Systems des Projekts KIT-FAST



Wilhelm Weisweber

 
July 1994, 142 pages
 
 
An experimental machine translation (MT) system has been developed and implemented within the project KIT-FAST, which translates written German texts into English. For that reason a syntactic, semantic and aspects of a conceptual level of representation have been realized. The syntactic and semantic representations are structures, which are generated by Generalized Phrase Structure Grammars (GPSG), and Functor-Argument-Structures (FAS), respectively. At the conceptual level aspects of background and textual knowledge of German are represented in the TBox and ABox of the knowledge representation system BACK that has been integrated into the MT system. These knowledge representations are needed for the evaluation of anaphoric relations in German texts.
 
The paper consists of three parts. In the first part the architecture of the MT system is outlined. Short descriptions of the representations (GPSG, FAS and text representation), of the algorithms (morphological analysis and synthesis, GPSG parser, interpreter for non-confluent term-rewrite systems and component for the evaluation of anaphoric relations) and of the structural mappings (morphological analysis and synthesis, syntactic, semantic and conceptual analysis, transfer and generation) are given. The second part is the installation guide of the MT system, which says how to get and install the MT system and shows the implementation of the interpreter for non-confluent term-rewrite systems, the unification algorithm, which is used by the interpreter, and the internal representation of the linguistic data. The third part is the user's guide of the MT system. The appendix includes the text fragment on which the linguistic data of the MT system is based.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 115


Zur Entstehung hierarchischer Kategoriestrukturen in einem selbstorganisierten neuronalen Netz



Daniel Durstewitz

 
March 1994, 25 pages
 
 
Various researchers have proposed that natural categories are in the first instance learned implicitly and without environmental feedback. The development of natural category systems is governed by economical principles which work on the basis of feature similarities between the objects. The advantage of such a category system is that unobserved features of objects and features of objects never seen before can be predicted reliably. The present article deals with a self-organizing neural net which is capable of forming hierarchical category structures in an unsupervised manner thereby using principles known from human category systems. As shown by simulation results the hierarchical levels of the network category system share important properties with the basic level, superordinate level and subordinate level of natural category systems. The neural net presented here is therefore suggested as a psychological model for unsupervised human category learning.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 114 < in German >


Zum Einfluß des Situationsmodells auf die Verarbeitung konsistenter versus inkonsistenter Textinformation



Daniel Durstewitz, Ute Schmid, Berry Claus, Klaus Eyferth

 
January 1994, 15 pages
 
 
Most current models of reading comprehension assume that readers generate at least two levels of representation of a text, a text-based representation and a situational or mental model. This is the basis for the experiment described here, too. We investigated the impact of mental models on text comprehension. Subjects read narratives describing scenes containing a character. Target objects were either spatially associated or spatially dissociated with the character. The last sentence of each narrative presented information either consistent or inconsistent with the target's location. We supposed that inconsistent information would influence text comprehension especially under the associated condition. Our results indicate that there is just a global impact of consistent vs. inconsistent information on text comprehension.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 113


Comparing O2 and BACK: A Case Study of Developing an Object Oriented Database Schema from a Medical Semantic Data Model



Eugene P. Woodfin

 
November 1993, 31 pages
 
 
Both object-oriented database systems and knowledge representation systems based on description logics address the need for dealing with complex, highly structured data in a simple, efficient, and flexible manner. In this paper, the object-oriented database O2 and BACK, a description logic based system, are compared in terms of expressiveness, flexibility of data access, and changeability. The comparison was performed by implementing an existing BACK model from a medical domain as an O2 schema providing roughly the same functionality.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 112


Persistent Maintenance of Object Descriptions using BACK



Albrecht Schmiedel

 
November 1993, 17 pages
 
 
A method for constructing and maintaining a `semantic index' using BACK, a system based on description logics, is described. A persistent index into a large number of objects is built by classifying the objects with respect to a set of indexing concepts and storing the resulting relation between object-ids and most specific indexing concepts on a file. The number of objects indexed is not limited by what BACK can hold in working memory, since the index files are incrementally updated. The index can be used for efficiently accessing the set of objects matching a query concept. The query is classified, and based on subsumption and disjointness reasoning with respect to indexing concepts instances are immediately categorized as hits, misses or candidates with respect to the query. Based on the index only, delayless feedback concerning the cardinality of the query (upper and lower bounds) can be provided during query editing.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 111


Deriving Inference Rules for Description Logics: a Rewriting Approach into Sequent Calculi



V. Royer, J. Joachim Quantz

 
December 1993, 125 pages
 
 
Description Logics (DL) can be investigated under different perspectives. The aim of this report is to provide the basis for a tighter combination of theoretical investigations with issues arising in the actual implementation of DL systems. We propose to use inference rules, derived via the Sequent Calculus, as a new method for specifying terminological inference algorithms. This approach combines the advantages of the tableaux methods and the normalize-compare algorithms that have been predominant in terminological proof theory so far. In our paper presented at JELIA'92 we proposed a generic method for deriving complete sets of inference rules for DL. The method relies upon translations into Sequent Calculus and systematic rewriting of sequent proofs. We illustrated our method on a relatively restricted terminological logic. In this report the approach is extended to the more expressive logic underlying Back V5. It turns out that concept-forming operators involving equality and role-forming operators considerably increase the complexity of our rewriting strategy.
 
The derived inference rules can be used in two ways for the characterization of DL systems: first, the incompleteness of systems can be documented by listing those rules that have not been implemented; second, the reasoning strategy can be described by specifying which rules are applied forward and which backward.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 110


Preferential Default Description Logics



J. Joachim Quantz and Mark Ryan

 
December 1993, 60 pages
 
 
This report is based on the default integration for Description Logics presented by Quantz and Royer at KR'92. We specify the general framework for Preferential Default Description Logics (PDDL), and present various alternative PDDL, addressing issues like implicit objects, priorization between defaults, and maximization of default application. We show that all PDDL satisfy the formal properties of system P (Kraus, Lehman and Magidor) and that tex2html_wrap_inline1375 , which is based on a priority ordering on multisets of defaults and motivated by an application in Natural Language Processing, even satisfies Rational Monotonicity. Finally, we show paralles between PDDL and Ryan's Ordered Theory Presentations.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 109


Task-Oriented Modeling for Natural Language Processing Systems



Gudrun Klose

 
September 1993, 150 pages
 
 
In this work, I focus on the crucial significance of task orientation as a design criterion for knowledge bases in natural language processing systems, a perspective which I realized with the implementation of the LEU/2 knowledge base for text understanding. The approach outlined in this thesis transcends current contributions in the field of Knowledge Modeling by offering a systematic account of modeling factors. On this basis, I give insights intodesign decisions and the access to knowledge elements during system performance by means of examples from three different natural language processing applications.
 
Starting with a discussion of the key role of task orientation at the intersection of cognitive science, epistemology, and computational linguistics as related disciplines, I continue with the outline of a differentiated task-based description scheme for knowledge base contents and functionalities. For this purpose, I draw on elements of a methodology originally developed within the field of Open Distributed Processing. The notions of aspects and views are tailored to theanalysis of knowledge bases in natural language processing systems. The resulting model of task orientation is validated by applying it to the respective knowledge bases of the LEU/2 system (IBM Scientific Center), the FAST machine translation system (Technical University Berlin), and the PENMAN natural language generation system (Information Sciences Institute, Los Angeles).
 
I conclude by sketching future perspectives for a constructive modeling methodology, and by characterizing ongoing research efforts concerning the investigated knowledge bases in terms of the introduced methodological aspects.
 

KIT-Report 108 < in German >


Anapherninterpretation in der Maschinellen Übersetzung (Schlußbericht des Berliner Projekts der EUROTRA-D-Begleitforschung)



Christa Hauenschild, Bernd Mahr, Susanne Preuß, Birte Schmitz, Carla Umbach, Wilhelm Weisweber, Lone Beheshty, Guido Dunker, Matthew Rickard, Christian Werner-Meier, Erich Ziegler

 
June 1993, 80 pages
 
 
The project dealt with the resolution of anaphoric expressions with respect to the need of Machine Translation. This involves several aspects that are crucial for Machine Translation: Translation of texts instead of single sentences, treatment of ambigous expressions, integration of background knowledge and strategies to deal with uncertain knowlegde. The project concentrated on personal and possessive pronouns that refer to objects and are referentially identical with their antecedents. A resolution procedure was developed that integrates different criteria (morphological, syntactic, semantic and conceptual) to determine the antecedent of a pronominal anaphor. The resolution procedure is based on a dual representation of the text: The structural aspects are represented by means of the Functor Argument Structure that has been developed in the preceeding project. The referential aspects are represented by the knowlegde representation system BACK that has been integrated into the Berlin MT-System. On the basis of the criteria so far developed remarkably good results were achieved. Moreover, the treatment of personal and possessive pronouns was unified. It is possible to apply the resolution procedure to other types of ambiguity as well, but then one has to take interdependencies between ambiguities into account. Besides anaphoric resolution other interfering subjects were dealt with, such as problems of consistency of the lexicon, of logical foundation of the MT system and of the term rewriting procedure that is used for various mappings within the system. A detailed description of the results can be found in the final project report.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 107


Inferences for Temporal Object Descriptions in a Terminological Representation System: Design and Implementation



Andreas Neuwirth

 
April 1993, 46 pages
 
 
This paper describes design and implementation of an assertional component (ABox) of a temporal terminological representation system. With this ABox, and the temporal TBox developed by Martin Fischer (see KIT-Report 99), it is possible to handle temporal object descriptions, and to infer 'new' (implicit) knowledge. An ABox query language offers a variety of possible queries to get certain information.
It is examined how existing algorithms for an ABox of a conventional terminological system can be transferred to the temporal system. Besides completely new algorithms have to be developed, because new problems arise with the temporal extension.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 106


Defaults in Machine Translation



Birte Schmitz, Joachim Quantz

 
February 1993, 31 pages
 
 
In this paper we try to combine experiences gained in the Machine Translation project FAST with research concerning the default extension of the terminological representation system BACK. We analyze various types of ambiguities that are problematic for translation and show that disambiguation is possible on the basis of default assumptions. Some of these defaults form preference rule systems, while others are merely an elegant way of modeling exceptions. We show how the ideas underlying the FAST implementation can be reformulated in a declarative framework by using terminological logics. In particular the implementation of anaphora resolution can only be expressed in this framework with resort to defaults. Whereas default reasoning is usually concerned with the problem of deducing valid extensions of a given set of formulas, preference rule systems pose slightly different requirements. A preference ordering of the possible extensions can be computed by taking into account which defaults are violated in them. Since the extension with the qualitatively minimal number of exceptions is the preferred one, disambiguation can be modeled by exception minimization.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 105


Retraction of Object Descriptions in BACK



Carsten Kindermann

 
December 1992, 41 pages
 
 
Terminological representation systems permit the construction of knowledge bases and schemata around the notion of concepts, roles, and instantiating objects. Inferential services they provide include checking for inconsistencies in and classification of concept and object descriptions. To improve their performance, several systems store derived propositions together with user-told data in the knowledge base. In this report we address the problem of retraction of object descriptions for systems that employ such a generative implementation. We adopt a data dependency network approach.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 104


Anaphora Resolution in Machine Translation



Susanne Preuß Birte Schmitz Christa Hauenschild Carla Umbach

 
December 1992, 24 pages
 
 
In this paper we give an overview of an approach to anaphora resolution that takes a whole variety of different factors into account. These factors concern on the one hand structural information like agreement, proximity, binding principles, subject preference, topic preference, negative preference for free adjuncts and on the other hand information about the contents of a text like conceptual consistency. This led to a twofold text representation: a structural text representation and a referential text representation. The factors are implemented as preference rules with different weights that express the influence each of the factors has in the process of anaphora resolution. For each factor a linguistic motivation as well as a formal representation of the information it works on is given. We show how we integrated the anaphora resolution component into the existing experimental MT system.
 

KIT-Report 103


How to Compute 1+1? A Proposal for the Integration of External Functions and Computed Roles into BACK



Gerd Kortüm

 
January 1993, 23 pages
 
 
Today, terminological knowledge representation systems have reached a level of maturity which makes their use for real world applications feasible. However, the construction of practical knowledge representation management systems (KBMS) requires more than a well-defined representation formalism and reasoning algorithm. Flexibility and system integration are two important issues for any KBMS. In this paper, we propose an extension of the BACK terminological representation system which allows the integration of externally realized procedures into the representation formalism. For that purpose, we introduce a new language construct External Function which serves as an interface to arbitrary host-language procedures. External Functions allow us to integrate forms of domain specific knowledge such as arithmetic computations into BACK which can not be expressed efficiently in other representation systems. We show how results of computations can be integrated into the reasoning process and how Computed Roles can be realized by means of external functions. As a result, we get an extensible knowledge representation system which primarily offers the advantages of a procedural representation formalism and can serve as a flexible tool for application development.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 102


Direct Parsing With Metarules



Wilhelm Weisweber, Susanne Preuß

 
December 1992, 12 pages
 
 
In this paper we argue for the direct application of metarules in the parsing process and introduce a slight restriction on metarules. This restriction relies on theoretical results about the termination of term-rewrite systems and does not reduce the expressive power of metarules as much as previous restrictions. We prove the termination for a set of metarules used in our German grammar, and show how metarules are directly interpreted by the Earley algorithm.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 101


Term-Rewriting as a Basis for a Uniform Architecture in Machine Translation



Wilhelm Weisweber

 
December 1992, 16 pages
 
 
In machine translation (MT) different levels of representation can be used to translate a source language sentence onto its target language equivalent. These levels have to be related to each other. This paper describes a declarative formalism on the basis of term-rewriting which maps one representation onto an equivalent adjacent one. The different levels (e.g. represented by derivational trees, feature structures or expressions of a knowledge representation language) can be represented as first order terms which are generated by signatures. The equivalences between them are stated as axioms which are directed to form a non-confluent and terminating term-rewrite system. A complete and coherent algorithm has been developed which interprets these systems, and is able to handle default rules.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 100


BACK V5 - Tutorial & Manual



Thomas Hoppe, Carsten Kindermann, Joachim Quantz, Albrecht Schmiedel, Martin Fischer

 
March 1993 , 135 pages
 
 
Since 1985 the terminological representation system BACK (Berlin advanced computational knowledge representation system) has been developed at the Technical University Berlin. Its origin lies in the KL-ONE-based knowledge representation paradigm, semantic networks, and frame-based representation languages. The current system version has gone through several changes not only on the implementational level, but also on the conceptual level. For some time now the syntax has stabilized and larger applications may be approached in the future.
 
Until now, introductions to the BACK system as well as its representation language were distributed throughout several publications, making it difficult for users to learn to handle the system. Hence, we found it worthwhile to write a tutorial guide through the BACK system. With BACK V5 the representation language has changed. On the one hand it became much more uniform than in previous versions, on the other hand it was extended by some useful constructs which allow an easier customization of the language. Thus, we felt also the need for a user manual documenting the revised language.
 
The first part of this report is a tutorial introduction into the BACK system which describes by example the modeling process we find most appropriate for terminological modeling with BACK. The second part is written in the form of a user manual, describing the modified and newly introduced language constructs.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 99


The Integration of Temporal Operators into a Terminological Representation System



Martin Fischer

 
December 1992, 50 pages
 
 
In this paper a temporal extension of terminological logics is provided. Operators from propositional temporal modal logic, tense logic, are added to a simple terminological language. With these operators knowledge related to sets of time points can be represented. Temporal concepts and patterns about past and future can be defined. Modifying and widening the tense-logical extension given by K. Schild (KIT-Report 92), I present a terminological and an assertional component and extend the language with ``When is ...?'' queries for both components. A model theoretic semantics for the overall formalism is given as well as correct but imcomplete structural algorithms for sublanguages. A temporal pattern interpreter to define and recognize arbitrary temporal patterns from temporal indexed data is provided too. Some remarks are made about the realization of the experimental temporal tex2html_wrap_inline1377 BACK system and about an intergration of tense logical operators into a real system like BACK as well.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 98 < in German >


EpsilonT - Eine Logik erster Stufe mit Selbstrefenz und totalem Wahrheitsprädikat



Werner Sträter

 
October 1992, 83 pages
 
 
Logics with a truth predicate on the object level have to face the problems of paradoxes. They may therefore either leave out self referent propositions (e.g. Russell Tarski), or use three or more truth values (e.g. Kleene) on the meta level, or the truth predicate on the object level does not coincide with that on meta level (e.g. Gupta, Aczel). In this paper, however, we present and investigate a two valued logic, which allows to express unlimited self reference, and in which the truth predicate of object and meta level coincide precisely (the Tarski biconditionals hold).
 

KIT-Report 97

Workshop on Text Representation and Domain Modelling Ideas from Linguistics and AI

Susanne Preuß, Birte Schmitz (eds.)
May 1992, 147 pages
Text representation and domain modelling are two topics that are of emerging interest for natural language processing (NLP) systems. Text representation has to be included in NLP systems if one wants to overcome the limitations of processing single sentences only and domain modelling is necessary in order to supply the system with additional information that is needed for the disambiguation of linguistically possible structures. In artificial intelligence (AI) some amount of work has been done on these topics but not necessarily with applications in NLP in mind. Therefore it was the aim of this workshop to bring together scientists from the fields of Linguistics and AI in order to discuss such topics as:
  • Is there a boundary between encyclopedic and linguistic knowledge? 
  • If we have such a boundary in our model, how do the encyclopedic and the linguistic levels interact? 
  • What kinds of knowledge bases do we need for special tasks in natural language processing, eg. machine translation? 
  • How can we build up a text representation? 
  • Are there any existing formalisms which are adequate for text representation?
The participants of the workshop came from the areas of machine translation, knowledge engineering, systemic grammar, terminological logic, architecture for integrated speech-language-systems and discourse representation theory.

KIT-Report 96 < in German >


Semantische Repräsentation anaphorischer Bezüge in terminologischen Logiken



Joachim Quantz

 
January 1992, 130 pages
 
 
Because of their referential ambiguity anaphoric expressions are considered as a central problem in natural language processing. Four basic types of anaphoric relations are distinguished: identity anaphora, contiguity anaphora, substitution anaphora, and bound anaphora. A more detailed classification of anaphora uses as criteria the lexical realization of the anaphor, the ontological category of its referent, the syntactic category of antecedent and anaphor, and the relation between antecedent and anaphor.
 
The use of terminological representation systems for the representation of discourse contents and background knowledge provides a solid basis for the resolution of anaphoric relations. In order to represent bound anaphora, however, generalized quantifiers have to be integrated into the framework of terminological logics.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 95


Terminological Logic Users Workshop - Proceedings -



Christof Peltason, Kai von Luck, Carsten Kindermann (eds.)

 
December 1991, 194 pages
 
 
The report contains the working papers submitted to the ``Terminological Logic Users Workshop'' held at Technical University Berlin on October 24-25, 1991 which was attended by more than 50 participants. The papers present application-oriented work using systems following the ``terminological'' (KL-ONE like) paradigm. The applications range from domain-specific examples in configuration and consultancy, to experience reports in document management. Applications of global data models for databases are discussed. The report also contains brief descriptions of the demos given and summaries of the discussions on special topics during the parallel sessions.
 

KIT-Report 94 < in German >


Anaphern-Interpretation in der Maschinellen Übersetzung



Christa Hauenschild

 
November 1991, 23 pages
 
 
The paper reports on research on anaphora interpretation in a machine-translation context. The relevance of anaphora interpretation for human as well as machine translation is argued for, an overview of relevant criteria for finding the intended antecedent is given for a subtype of anaphoric pronouns, and the interaction of different types of ambiguities (referential, lexical and ``relational'') in finding a consistent interpretation hypothesis is discussed. The actual state of modelling the relevant factors in our machine-translation system is presented, and, finally it is argued why the research reported on in the paper can be considered a contribution to the development of a cognitive translation theory.
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 93 < in German >


Textrepräsentation und Hintergrundwissen für die Anaphernresolution im Maschinellen Übersetzungssystem KIT-FAST



Birte Schmitz, Susanne Preuß, Christa Hauenschild

 
September 1991, 37 pages
 
 
n this paper we want to show how representation of textual and background knowledge can be used fo Machine Translation (MT). As a basis we took the result of the KIT FAST Project, which started from the problem of anaphora resolution. We argue that two kinds of text representation are useful in order to deal with intersentential phenomena: one is to represent structural characteristics of the text, and the other is to represent aspects of the contents. We discuss a concept of modeling background knowledge for MT, that is based on task-oriented criteria. On balance the result is that MT and especially anaphora resolution require a language particular modeling of conceptual knowledge. As a means to represent text contents and background knowledge the project KIT FAST uses the BACK-System, that is based on terminological logic.
 

KIT-Report 92


A Tense-Logical Extension of Terminological Logics



Klaus Schild

 
September 1991, 16 pages
 
 
A temporal extension of terminological logics based on the well-known point-based temporal operators of tense logic is presented. It is shown that the tense-logical extension of propositionally closed terminological logics is fully expressive for integer and real time. That is, each first order temporal operator over integers and real numbers can be defined within the tense-logical extension based on integer and real time, respectively. Moreover, it is proved that, for integer time, the tense-logical extension of the propositionally closed concept language ACL has the same computational complexity as its base ACL . Hence, for point-based languages over integer time, the presented temporal extension constitutes a best possible tradeoff between expressiveness and computational complexity.
 

KIT-Report 91


A Correspondence Theory for Terminological Logics:
- Preliminary Report -



Klaus Schild

 
August 1991, 14 pages
 
 
We show that the terminological logic ALC comprising Boolean operations on concepts and value restrictions is a notational variant of the propositional modal logic K(m) . To demonstrate the utility of the correspondence, we give two of its immediate by-products. Namely, we axiomatize ALC and give a simple proof that subsumption in ALC is PSPACE-complete, replacing the original six-page one.
 
Furthermore, we consider an extension of ALC additionally containing both the identity role and the composition, union, transitive-reflexive closure, range restriction, and inverse of roles. It turns out that this language, called TSL , is a notational variant of the propositional dynamic logic converse-PDL. Using this correspondence, we prove that it suffices to consider finite TSL -models, show that TSL -subsumption is decidable, and obtain an axiomatization of TSL .
 
By discovering that features correspond to deterministic programs in dynamic logic, we show that adding them to TSL preserves decidability, although violates its finite model property. Additionally, we describe an algorithm for deciding the coherence of inverse-free TSL -concepts with features. Finally, we prove that universal implications can be expressed within TSL .
 
To download the pdf version of this report click below:
Download PDF Version.
To download the ps version of this report click below:
Download PS Version.

KIT-Report 90


Semantic Validation of Inferences in Terminological Representation Systems



Jan Thomsen

 
August 1991, 43 pages
 
 
Based on the semantics of a BACK-like Term Subsumption Language we give an overview of different settings of terminologies based on a terminological language including number restrictions, value restrictions, disjointness and individual restrictions, and primitive subroles, discussing the inferences which have to be drawn in them (TBox inferences). The same is done by considering the inferential consequences of a rule operator (IBox inferences) and a basic assertional language (ABox inferences). By this we get a number of little abstract examples which serve as the description of an environment for testing Terminological Representation Systems semantically.
 

KIT-Report 89


International Workshop on Terminological Logics - Proceedings -



Bernhard Nebel, Christof Peltason, Kai von Luck (eds.)

 
August 1991, 131 pages
 
 
The report is a compilation of working papers submitted to the ``Second International Workshop on Terminological Logics'' held at Schloß Dagstuhl near Saarbrücken, Germany, on May 6-8, 1991. The workshop brought together 40 invited participants currently working in the field, and served to provide a snapshot of the current state of research. As documented by the variety of talks, the aspects of theoretical work (semantical foundations, complexity), system-oriented work (implementations), and application-oriented work are all dealt with within one community. In character with the informal nature of the workshop, these papers sketch personal interests, work in progress, or summaries of research results rather than being fully elaborated articles.
 

KIT-Report 88 < in German >


Schlußbericht des Berliner Projekts der EUROTRA-D- Begleitforschung ``Transfer und Generierung auf satzsemantischer Basis''



Stephan Busemann, Tim Eckardt, Margaret Garry, Christa Hauenschild, Bernd Mahr, Ann McLarnon, Susanne Preuß, Birte Schmitz, Carla Umbach, Wilhelm Weisweber, Christian Werner-Meier, Lucy Wilson, Erich Ziegler

 
Juni 1991, 41 pages
 
 
The research of the Berlin project of the complementary research for EUROTRA-D is based on a model of a translation system, which assigns for four levels of representation: a syntactic and a sentence semantic level, a level for the representation of texts and as an invariant the thematic and argumentative structure of text. In the preceding project ''New algorithms for analysis and synthesis for machine translationthe Generalized Phrase Structure Grammar (GPSG) has been investigated for its applicability for the syntactic level of representation and has been implemented to verify the hypotheses, which have been developed.
 

Zusatzinformationen / Extras

Direktzugang

Schnellnavigation zur Seite über Nummerneingabe