103 research outputs found
The Role of Technology in Distributed Team Coordination: A multi-method investigation of a technology change process in the field
No full textThe goal of this project was two-fold: firstly, to add to the theoretical knowledge about the impact of complex media combinations for distributed team coordination, and secondly, to increase the practical knowledge for organizations and designers on how to design and implement complex technological solutions for the support of distributed teams. In this thesis I report the results of six empirical studies conducted in offshore oil production teams. The main theoretical contributions lay in the introduction of the concept of asymmetry and its effects on distributed team work, the reframing of IT implementation and adoption as a team-level process, the development of a model for IT adoption in distributed teams, a delineation of effects of technological versus structural means to bridge geographic distribution, and a detailed view on the processes, with which multiple technologies impact on team functioning and more specifically intra-team coordination in mature distributed teams. For organizations, the knowledge gained from this project can help anticipate effects of media choices on their distributed work force and guide decisions of technology choice and design. The valence–alignment framework introduced in this thesis further provides managers and organizations with a framework to analyze reactions of user to technology implementations and devise strategies against resistance. For designers this projects aims to raise awareness for the relevance of social inter-dependencies and dynamics for product acceptance and use, the possibility of disparate or even conflicting user requirements in multi-user contexts, and the importance of a stronger process view from initial attitudes to long-term use.Product Innovation ManagmentIndustrial Design Engineerin
An exploration of knowledge integration problems in interdisciplinary research teams
Get PDFThe integration of function-specific expertise into a shared knowledge base is a crucial, but complex process for success in interdisciplinary teams. This paper presents an empirically derived typology of knowledge integration problems and links their occurrence to degree of heterogeneity and present stage of a team’s life.Industrial Design Engineerin
Technology as enabler for empowerment in distributed teams - a field study on leadership attitudes.
No full textIn this field study on distributed teams we examined the impact of communication media on attitudes towards empowerment and integration of remote subgroups. Using Q-methodology, interviews and questionnaires we compared attitudes of team members and managers in low and high media-rich environments. Our results indicate that use of richer media was connected with higher appreciation of team empowerment and subgroup integration, but only in groups in consistent contact with the media. Interview data further indicated that changed attitudes coincided with actual changes in leadership behaviors. Implications for leadership in distributed teams are discussed.Product Innovation ManagementIndustrial Design Engineerin
bacon: Linked Data Integration based on the RDF Data Cube Vocabulary
No full textDiscovering and integrating relevant real-live datasets are es-
sential tasks, when it comes to handling Linked Data. Sim-
ilar to Data Warehousing approaches, Linked Data can be
prepared to enable sophisticated data analysis. The devel-
oped open source framework
bacon
enables interactive and
crowed-sourced Data Integration on Linked Data (Linked
Data Integration), utilizing the RDF Data Cube Vocabulary
and the semantic properties of Linked Open Data. Discov-
ering suitable datasets on-the- y in local or remote repos-
itories sets up the ensuing integration process. Based on
well-known Data Warehousing processes, the semantic na-
ture of the data is taken into account to handle and merge
RDF Data Cubes. To do so, structure and content of the
cubes must be analyzed and processed. A similarity measure
has been developed to nd similarly structured cubes. The
user is o ered a graphical interface, where he can search for
suitable cubes and modify their structure based on semantic
properties. This process is fostered by a set of automated
suggestions to support inexperienced users and also domain
expert
A Comparison of Hybrid and End-to-End Models for Syllable Recognition
No full textThis paper presents a comparison of a traditional hybrid speech recognition system (kaldi using WFST and TDNN with lattice-free MMI) and a lexicon-free end-to-end (TensorFlow implementation of multi-layer LSTM with CTC training) models for German syllable recognition on the Verbmobil corpus. The results show that explicitly modeling prior knowledge is still valuable in building recognition systems. With a strong language model (LM) based on syllables, the structured approach significantly outperforms the end-to-end model. The best word error rate (WER) regarding syllables was achieved using kaldi with a 4-gram LM, modeling all syllables observed in the training set. It achieved 10.0% WER w.r.t. the syllables, compared to the end-to-end approach where the best WER was 27.53%. The work presented here has implications for building future recognition systems that operate independent of a large vocabulary, as typically used in a tasks such as recognition of syllabic or agglutinative languages, out-of-vocabulary techniques, keyword search indexing and medical speech processing
Maschinelles Lernen zur Detektion und Klassifikaton von Stottern in der Stottertherapie
Get PDFThis thesis comprehensively explores stuttering detection and classification, lever- aging diverse datasets and methodologies. It emphasizes the importance of stuttering classification in enhancing accessibility for individuals with speech disorders and delivers a structured overview, starting with the foundational elements of speech and machine learning and extending to the new methods developed during the course of this research.
The thesis identifies existing research gaps, necessitating a thorough examination of available datasets and a review of historical methods deployed for stuttering detection. It highlights the absence of comparability in the field due to diversity in annotation methods and the scarcity of datasets.
This work is a crucial step in advancing the automation of stuttering assessment and therapy evaluation. It introduces and evaluates the speech control index, a new metric to assess stuttering therapy recordings. It goes on to create the Kassel State of Fluency (KSoF) dataset, advancing German stuttering research with optimized annotation protocols and enabling cross-language research initiatives. An in-depth analysis of the Stuttering Events in Podcasts (SEP-28k) brings forth valuable insights into its composition, advocating for quality assurance in research data and speaker exclusivity while splitting data into train, development, and test sets.
This thesis investigates speech transformer features for stuttering detection and sets new benchmarks in stuttering classification. It conclusively shows the utility of features learned using English stuttering data on German stuttering recordings. Furthermore, it evaluates end-to-end stuttering classification systems based on speech transformer models. This is done using multi-language and cross-corpus datasets, showing the developed methods’ generalizability and their contribution towards a general stuttering detection system. It does so by evaluating cross-corpus and multi-language stuttering classification systems. While the research concludes with the harder task of multi-label stuttering classification, it underscores the ongoing challenges due to data availability and the inherent ambiguity of stuttering.
It advocates for the creation of new multimodal datasets and refinement in the processing of stuttered speech, focusing on improving the accessibility of speech technology. The thesis contains innovative contributions, empirical explorations, and insights, aiming to navigate the future pathway in the field of stuttering detection and classification.Diese Dissertation befasst sich umfassend mit der Erkennung von Stottern, wobei ver- schiedene Datensätze und Methoden genutzt werden. Sie unterstreicht die Bedeutung der Stotterklassifikation für die Verbesserung der Zugänglichkeit für Menschen mit Sprechstörungen und liefert einen strukturierten Überblick, der mit einer grundlegenden Betrachtung von Sprache und des maschinellen Lernens beginnt und sich auf die neuen Methoden erstreckt, die im Laufe dieser Forschung entwickelt wurden. Basierend auf einer gründlichen Untersuchung der verfügbaren Datensätze und einem Überblick über die Methoden, die in der Vergangenheit zur Stottererkennung eingesetzten wurden, werden Forschungslücken aufgezeigt. Dabei wird klar, dass mangelnde Vergleichbarkeit von Methoden aufgrund der Heterogenität der Datensätze ein Kernproblem ist.
Diese Dissertation ist daher ein entscheidender Schritt, um die Automatisierung der Beurteilung des Stotterns und der automatisierten Bewertung der Therapie vo- ranzutreiben. Sie führt den Speech Control Index ein, welcher eine neue Metrik zur Beurteilung von Stottertherapieaufnahmen ist. Darüber hinaus wird der Kassel State of Fluency (KSoF) Datensatz erstellt, der dazu beiträgt, die Stottererkennung im Thera- piebereich voranzutreiben. Mit dem Datensatz werden sprachübergreifende Experimente ermöglicht. Eine eingehende Analyse des Stuttering Events in Podcasts Datensatzes
(SEP-28k) liefert wertvolle Erkenntnisse über seine Zusammensetzung und plädiert für Qualitätssicherung bei Forschungsdaten und Sprechexklusivität bei der Aufteilung von Daten in Training-, Development- und Testpartitionen.
Diese Arbeit untersucht Transformer basierte Merkmale für die Stottererkennung und setzt damit neue Maßstäbe in der Stotterklassifikation. Sie zeigt schlüssig den Nutzen von Merkmalen, die mit englischen Daten gelernt wurden, für die Klassifikation von Stottern in deutschsprachigen Aufnahmen. Darüber hinaus werden End-to-End Klassi- fikationssysteme für Stottern auf der Grundlage von Transformermodellen erstellt und evaluiert. Dies geschieht mit mehrsprachigen und korpusübergreifenden Datensätzen, was die Generalisierbarkeit der entwickelten Methoden und ihren Beitrag zu einem generellen Stottererkennungssystem zeigt. Die Dissertation schließt mit der schwierigeren Aufgabe der Multi-Label-Klassifikation von Stottern und unterstreicht die Herausforderungen aufgrund der Datenverfügbarkeit und der Mehrdeutigkeit des Stotterns.
Diese Arbeit plädiert für die Schaffung neuer multimodaler Datensätze und die Verbesserung der Verarbeitung von gestotterter Sprache, wobei der Schwerpunkt auf der Verbesserung der Zugänglichkeit von Sprachtechnologie liegt. Die Dissertation enthält innovative Beiträge, empirische Untersuchungen und Erkenntnisse, die den zukünftigen Weg im Bereich der Stottererkennung und -klassifizierung aufzeigen sollen
- …
