05 Fakultät Informatik, Elektrotechnik und Informationstechnik

Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/6

Browse

Filters

Settings

Institute: search.filters.UBSInstitut.Institut für Maschinelle SprachverarbeitungEnd date: 2011

Search Results

Now showing 1 - 10 of 28
  • Thumbnail Image
    ItemOpen Access
    Fehlerbehandlung in Mensch-Maschine-Dialogen
    (2007) Gieselmann, Petra; Rohrer, Christian (Prof.)
    Seit es Computer gibt, existiert auch der Wunsch des Menschen, mit ihnen reden zu können wie mit einem anderen Menschen. Eines der berühmtesten Beispiele dafür ist sicherlich Eliza, ein Computerprogramm, das einen Psychologen simuliert, mit dem der Benutzer ein Therapiegespräch führen kann. In vielen Science-Fiction-Filmen finden sich auch immer wieder Beispiele für solche sprechenden Maschinen, wie beispielsweise HAL in 2001: Odysee im Weltraum'' oder auch der Computer auf dem Raumschiff Enterprise''. So reichen erste Dialogsysteme bereits zurück bis in die Anfänge der künstlichen Intelligenz in den fünfziger Jahren. Dennoch hatten diese Dialogsysteme bis vor wenigen Jahren noch mit so vielen Problemen zu kämpfen, dass sie kaum für einen praktischen Einsatz geeignet waren. Erst in letzter Zeit ist es durch die stetigen Verbesserungen im Bereich von Spracherkennung und Sprachverstehen und das Aufkommen von immer schnelleren und mächtigeren Rechnern möglich geworden, solche Systeme für den realen Einsatz zu bauen. Nach wie vor gibt es aber noch eine ganze Reihe ungelöster Probleme, die zum einen auf die Komplexität natürlicher Sprache und zum anderen auf den immensen Fundus an vernetztem Weltwissen und Kontextbeziehungen, über den Menschen verfügen, zurückzuführen sind. Eine der bislang größten Herausforderungen liegt darin, ein solches Dialogsystem auch für den realen Einsatz unter Alltagsbedingungen zu entwerfen. Bisher fehlt den Systemen dafür noch die nötige Fehlerrobustheit, um in Situationen, in denen das System etwas falsch verstanden hat und es zu Problemen kommt, angemessen reagieren zu können. In dieser Arbeit geht es genau um solche Fehler im Dialog, wie sie vermieden und während des laufenden Dialogs wieder behoben werden können, wenn sie nicht vorher zu vermeiden waren. Der Gegenstand dieser Arbeit ist eine datengetriebene Analyse der Fehler, die in der Mensch-Roboter-Kommunikation auftreten mit dem Ziel, diese möglichst im Vorfeld zu vermeiden. Es wird eine Fehlerklassifikation aufgestellt und es werden Methoden für die Vermeidung der verschiedenen Fehlerklassen entwickelt und evaluiert. Darüberhinaus werden auch generische Methoden zur Fehlerbehebung für die Fälle implementiert, die nicht vorher vermieden werden konnten, ebenfalls mit Hilfe datengetriebener Analysen. Damit soll es ermöglicht werden, Dialogsysteme über die Laborumgebung hinaus in realen Situationen einsetzen zu können. Dies wird am Beispiel eines Haushaltsroboters diskutiert und evaluiert. Diese Ausarbeitung gliedert sich in vier Teile: Der erste Teil beschäftigt sich mit dem Stand der Forschung in den Bereichen, die hier eine Rolle spielen. Dazu werden verschiedene Ansätze für Mensch-Maschine-Dialogsysteme beleuchtet. Im Anschluss wird die menschliche Informationsverarbeitung im Dialogbereich erläutert. Dabei geht es auch um Fehlerdialoge in zwischenmenschlichen Dialogen, die hier als Vorbild für Mensch-Roboter-Dialoge dienen. Der zweite Teil beschäftigt sich mit den durchgeführten Benutzertests und Datensammlungen und der Klassifikation von Fehlern im Dialog, die die Grundlage für die folgenden Arbeiten zur Fehlervermeidung und -behebung bilden. Zunächst erfolgt eine detaillierte Analyse von Fehlern, die bei der Mensch-Roboter-Interaktion auftreten können. Dazu werden verschiedene aufeinander aufbauende Benutzerstudien und Datensammlungen, bei denen der Roboter dem Menschen im Haushalt zur Hand geht und einfache Tätigkeiten verrichtet, durchgeführt, um eine große Menge an möglichst realistischen Daten gewinnen zu können, die nicht nur unter Laborbedingungen entstanden sind. Im dritten Teil werden verschiedene Methoden zur Fehlervermeidung und -behebung vorgestellt. Zur Fehlervermeidung werden zusätzliche Wissensquellen in den Dialogmanager integriert. Außerdem werden Mechanismen zur Anaphernresolution, Kontextmodellierung, Auflösung von Ellipsen, multimodalen Fusion und zum Umgang mit komplexen, zusammengesetzten Äußerungen entwickelt und evaluiert. Zur Fehlerbehebung werden verschiedene Strategien für effektive Klärungsfragen untersucht. Metakommunikation, wie sie in den durchgeführten Benutzertests vorkommt, wird analysiert, um eine effektivere Kommunikation gewährleisten zu können. Außerdem wird ein Mechanismus entwickelt, der es dem Roboter erlaubt, problematische Situationen zu erkennen und diese selbst durch Metakommunikation aufzulösen. Im vierten Teil werden die entwickelten Methoden anhand eines abschließenden Benutzertests evaluiert. Dabei geht es darum, das System mit allen entwickelten Mechanismen zur Fehlerbehandlung zu testen und es mit dem Basissystem zu vergleichen. Das besondere Augenmerk liegt hier auf der Übertragbarkeit der entwickelten Mechanismen auf andere Domänen und Systeme. Danach folgt das Fazit der gesamten Arbeit und eine Diskussion der zukünftigen Arbeiten im Hinblick auf mögliche Erweiterungen dieses Systems.
  • Thumbnail Image
    ItemOpen Access
    Methods for coreference visualization and annotation
    (2010) Burkovski, Andre; Heidemann, Gunther; Kobdani, Hamidreza; Schütze, Hinrich
    In this report we present methods for visualization of coreferences. We use Self Organizing Maps, graph-based models, and text-based techniques to visualize coreference information from a pair-wise model and offer an interactive presentation of the feature space. We introduce our software system and several visualizations of the Self Organizing Map. The visualizations enable the user to explore the feature space, provide insights for feature engineering and allow a fast annotation of data with coreference information.
  • Thumbnail Image
    ItemOpen Access
    Disambiguation and reambiguation
    (2009) Hamm, Fritz; Kamp, Hans; Solstad, Torgrim; Roßdeutscher, Antje (ed.)
    The papers in this volume developed as part of the two projects "The Role of Lexical Information in the Context of Word-formation, Sentence and Discourse" and the project "Representation of Ambiguities and their Resolution in Context". In the former, a theory of "-ung"-nominalisation in German has been developed. The two papers presented in this volume focus on the second part of the joint enterprise of the two projects, namely on disambiguation of "-ung"-nouns in context. Hamm and Kamp study a proto-typical example, "die Absperrung der Botschaft" "the cordoning-off of the embassy", which is three-way ambiguous. This DP can denote a material object (the fence used for cordoning-off), an event (the process of cordoning-off) or a result state (the embassy being cordoned off). Formally, this three-way ambiguity is represented by an underspecified DRS. The paper contributes a partial answer to the general question which contextual factors are responsible for the (partial) disambiguation of this DP in discourse. The disambiguation process is described on the level of DRT. Building on the results in the first paper, the second paper by Hamm and Solstad focuses on problems that arise in anaphora resolution of pronouns with ambiguous nouns like "die Absperrung der Botschaft" as antecedent. What happens if the selection restriction of the verb in the antecedent sentence and that of the consequent sentence are incompatible? This situation is exemplified in (1): (1) Die Absperrung der Botschaft wurde vorgestern von Demonstranten behindert. Wegen anhaltender Unruhen wird SIE auch heute aufrecht erhalten. "The cordoning-off of the embassy was hampered by protesters the day before yesterday. Due to continuing unrest, it [the state of being cordoned off] is sustained today as well." "Behindern" "to hamper" filters out both the entity-reading and the result state reading of "Absperrung", but the verb "aufrecht erhalten" "to sustain" requires the result state as its argument. Thus, in order for the anaphoric pronoun 'sie' to be resolved successfully, the first sentence should provide a result state which, however, is not available, if the result state reading has been erased. Hamm and Solstad show that the required result state can be reconstructed - even under the assumption that "behindern" erases the result state reading of the first sentence in (1). This is achieved in a process of "reambiguation". Reambiguation involves a non-monotonic inference process. The question arise what triggers this process and what its restrictions are. Hamm and Solstad provide formally precise answers to these questions. Again a combination of UDRT and the event calculus provide the framework where these puzzles can be solved.
  • Thumbnail Image
    ItemOpen Access
    Semantic and pragmatic aspects of some particular uses of contrast marking
    (2006) Soffner, Martin; Bäuerle, Rainer (apl. Prof. Dr.)
    The aim of the thesis is an analysis of the relational meaning of contrast, as realised through the English co-ordinating conjunction "but". Since the analysis is to be based on well-defined contexts, first the essential properties of some selected utterance contexts of "but" have to be described. Then we seek after interrelations between "but" and those properties. In a stricter sense, this thesis is a case study of the use of "but" in answer situations: It is a question which defines the utterance contexts of "but" being considered. The definition of these dialogical contexts is a central concern. A question semantics that seems adequate for this purpose is one which determines questions by way of their possible answers (Groenendijk & Stokhof 1984). In order for an utterance to count as an answer, it has to meet some constraints that are due to the question. In particular, a question context provides a question domain that the answerer has to take into account. The impact of such a domain on the interpretation of utterances is an important aspect. What role might contrast play in a question context in which the exhaustivity condition must be fulfilled? We then turn to a notion of context extended by a superordinate problem (issue). Issues are a speaker's motive for asking a question. There are some descriptive concepts that model this motive for utterances, e.g. Conversational Topic (van Kuppevelt 1996), Question under Discussion (Roberts 1996), and Decision Problem (van Rooy 2003b). A simple model of issue is introduced that involves another kind of domain which differs from the question domain. This kind of domain can be distinguished and described by means of a semantics of counterfactuals (Kratzer 1981a). The notion of perspective is introduced; it is capable of subsuming both kinds of domains, consisting of individuals or of propositions focused by the speaker. The use of "but" also interrelates with the issue-related domain: Contrast can be explained by a shift of perspective; the domain currently focused by the speaker changes in a characteristic way. In a felicitous discourse, the answer can be expected to be aligned with the issue behind the question. If "but" indicates a change of the question domain, then this should also reflect a change of the decision of the issue. So contrast can be said to interrelate with the perspective chosen by the speaker.
  • Thumbnail Image
    ItemOpen Access
    The perfect time span : on the present perfect in German, Swedish and English
    (2006) Rothstein, Björn Michael; Kamp, Hans (Prof. Dr. h.c. PhD)
    This study proposes a discourse based approach to the present perfect in German, Swedish and English. It is argued that the present perfect is best analysed by applying an ExtendedNow-approach. It introduces a perfect time span in which the event time expressed by the present perfect is contained. The present perfects in these languages differ with respect to the boundaries of perfect time span. In English, the right boundary is identical to the point of speech, in Swedish it can be either at or after the moment of speech and in German it can also be before the moment of speech. The left boundary is unspecified. The right boundary is set by context.
  • Thumbnail Image
    ItemOpen Access
    German clause-embedding predicates : an extraction and classification approach
    (2010) Lapshinova-Koltunski, Ekaterina; Heid, Ulrich (Prof. Dr. phil. habil.)
    This thesis describes a semi-automatic approach to the analysis of subcategorisation properties of verbal, nominal and multiword predicates in German. We semi-automatically classify predicates according to their subcategorisation properties by means of extracting them from German corpora along with their complements. In this work, we concentrate exclusively on sentential complements, such as dass, ob and w-clauses, although our methods can be also applied for other complement types. Our aim is not only to extract and classify predicates but also to compare subcategorisation properties of morphologically related predicates, such as verbs and their nominalisations. It is usually assumed that subcategorisation properties of nominalisations are taken over from their underlying verbs. However, our tests show that there exist different types of relations between them. Thus, we review subcategorisation properties of morphologically related words and analyse their correspondences and differences. For this purpose, we elaborate a set of semi-automatic procedures, which allow us not only to classify extracted units according to their subcategorisation properties, but also to compare the properties of verbs and their nominalisations, which occur both freely in corpora and within a multiword expression. The lexical data are created to serve symbolic NLP, especially large symbolic grammars for deep processing, such as HPSG or LFG, cf. work in the LinGO project (Copestake et al. 2004) and the Pargram project (Butt et al. 2002). HPSG and LFG need detailed linguistic knowledge. Besides that, subcategorisation iformation can be applied in applications for IE, cf. (Surdeanu et al. 2003). Moreover, this information is necessary for linguistic, lexicographic, SLA and translation work. Our extraction and classification procedures are precision-oriented, which means that we focus on high accuracy of our extraction and classification results. High precision is opposed to completeness, which is compensated by the application of extraction procedures on larger corpora.
  • Thumbnail Image
    ItemOpen Access
    Phonetic and phonological development of stress in German
    (2010) Lintfert, Britta; Möbius, Bernd (Prof. Dr.)
    The purpose of this thesis is to describe the development of stress in children's speech and its evolution over the first few years of their lives. Within the results of the conducted analyses an exemplar-based speech production model of the acquisition of stress was developed. To this end the phonological and phonetic development of stress were analysed as well as the development of vowel space. The development of different perceptual spaces was adressed in the context of stress development of German children. Longitudinal speech samples of children from 5 months up to 36 months of age were recorded and analysed to create a prosodically annotated speech corpus of babbling, first words and meaningful speech of German-speaking children. In order to compare the child's use of the acoustic parametres, the parental use of the acoustic parametre of stress was also annotated and analysed. Furthermore, for describing the vowel development, cross-sectional data for children between 4 and 10 years were evaluated. First the crucial role of stress in the development of different syllable structures is described. Then the acoustic realisation of stress is analysed. The correlates of German word stress which include vowel duration, RMS Intensity, fundamental frequency, first four formants and different voice quality parameters, were therefore also analysed in children between 5 and 36 months of age and their parents. Besides, the development of the F1-F2 and F2-F3 vowel space depending on stress is described illustrating a cognitive map of representations in the perceptual space The development of the articulatory-acoustic relationship seems to be nonlinear and dependent on an increased control of motor abilities. The results of the different analyses are integrated in a comprehensive neural model of speech production using a self-organising neural network. This model is very appropriate to describe the differences in the production of babbling and words as there are possibilities to develop different neural maps depending on the stage of development as well as on the articulatory abilities. Self-feedback as well as feedback from the environment is also integrated in this model of production and perception.
  • Thumbnail Image
    ItemOpen Access
    User concepts for in-car speech dialogue systems and their integration into a multimodal human-machine interface
    (2010) Mann, Sandra; Dogil, Grzegorz (Prof. Dr. phil. habil.)
    In the automotive environment an increasing number of assisting systems as well as luxury features has found its way into vehicles over the past years: radio, CD, MP3, DVD, on-board address book, organisers, mobile phones, PDAs, destination input, selecting points of interest etc. The variety of on-board electronic devices and its extensive functions are likely to distract drivers from pursuing the primary driving task. Speech as a means of human-machine interaction should offer the user hands-free operation while driving to concentrate on the traffic situation rather than using various displays and buttons. However, parts of in-car dialogue systems tend to have lost transparency due to the increased functionality. Although people have accepted the innovative character of speech as interaction means and are generally motivated to use it, they fairly often get frustrated when operating an in-car speech dialogue system. Interaction is considered difficult and cumbersome and the first hurdle for getting acquainted with the system is high. It is often the conceptual design as opposed to technical mistakes that is regarded as the major obstacle for efficient and user-friendly interaction. The aim of this approach therefore was to develop concepts that facilitate human-machine interaction. The concepts range from general design guidelines to accessing large databases within applications such as audio, navigation and address book. Current methods of navigating data once were focused on very small numbers of music titles and address book entries and are therefore no longer adequate. To establish general requirements for the user interface, human communication was analysed to extract principles crucial for human-machine interaction. Also, various user studies were carried out to find out how users behave in human-computer interaction and where usability problems occur. The user's lacking recollection of audio, navigation and address book data was particularly taken into account and integrated into the concepts. The concepts bring back transparency into the multitude of technical devices and large amounts of data, while simultaneously ensuring consistency within and across modalities.
  • Thumbnail Image
    ItemOpen Access
    Segmental factors in language proficiency : degree of velarization, coarticulatory resistance and vowel formant frequency distribution as a signature of talent
    (2011) Baumotte, Henrike; Dogil, Grzegorz (Prof. Dr.)
    The present PhD proposes a reason for German native speakers of various proficiency levels and multiple English varieties producing their L2 English with different degrees of a foreign accent. The author took into account phonetic measurements to investigate the degree of velarization and coarticulation or coarticulatory resistance respectively in German and English, taking non-words and natural language stimuli. To get an impression of the differences between the productions of proficient, average and less proficient speakers in German and English, the mean F2 and Fv values in /ə/ before /l/ and in /l/ were calculated, for then comparing the degree of velarization in /əlV/ non-word sequences with each other. Proficient speakers gained lower formant frequencies for F2 and Fv in /ə/ than less proficient speakers, i.e. proficient speakers velarized more than less proficient speakers. Within the comparisons with respect to coarticulation or coarticulatory resistance results respectively the difference values for F2 and F2' out of /ə/ in /əleɪ/ vs. /əlu:/, /əly/ vs. /əleɪ/ and /əly/ vs. /əlaɪ/ were created. In the whole series of measurements, an overwhelming trend for proficient speakers being more coarticulatory resistant, i.e. velarizing more, and more precisely pronouncing English vowel characteristics than less proficient speakers was present, while average speakers did not continuously behave according to prediction, as a result of being sometimes “worse” than less proficient speakers. On the basis of Díaz et al. (2008) who pled for pre-existing individual differences in phonetic discrimination ability which enormously influence the achievement of a foreign sound system, it is claimed for a derivation of foreign language from native phonetic abilities.
  • Thumbnail Image
    ItemOpen Access
    Brain, meaning, and computation
    (2007) Klein, Michael; Kamp, Hans (Prof. Dr. h.c. PhD)
    This thesis deals with the question how the human brain acquires, represents, and processes the meaning of natural language expressions. A computational neural theory of meaning is introduced with the goal of overcoming the strong prevalence of empirical results over theoretical understanding that is currently present in the neuroscience of language. In this context, the brain is regarded as a goal-directed system, which acquires language and meaning as one means for achieving its goals. To accomplish complex learning tasks, such as acquiring a language, the brain uses subsystems, which differ especially with respect to their learning strategies, but interact so as to achieve the global goals of the system.