Bootstrapping techniques have the potential to accelerate the development of language technology resources. This is of specific importance in the developing world where language technology resources are scarce and linguistic diversity is high. In this thesis we analyse the pronunciation modelling task within a bootstrapping framework, as a case study in the bootstrapping of language technology resources. We analyse the grapheme-to-phoneme conversion task in the search for a grapheme-to-phoneme conversion algorithm that can be utilised during bootstrapping. We experiment with enhancements to the Dynamically Expanding Context algorithm and develop a new algorithm for grapheme-tophoneme rule extraction (Default & Refine) that utilises the concept of a ‘default phoneme’ to create a cascade of increasingly specialised rules. This algorithm displays a number of attractive properties including rapid learning, language independence, good asymptotic accuracy, robustness to noise, and the production of a compact rule set. In order to have greater flexibility with regard to the various heuristic choices made during rewrite rule extraction, we define a new theoretical framework for analysing instance-based learning of rewrite rule sets. We define the concept of minimal representation graphs, and discuss the utility of these graphs in obtaining the smallest possible rule set describing a given set of discrete training data. We develop an approach for the interactive creation of pronunciation models via bootstrapping, and implement this approach in a system that integrates various of the analysed grapheme-to-phoneme alignment and conversion algorithms. The focus of this work is on combining machine learning and human intervention in such a way as to minimise the amount of human effort required during bootstrapping, and a generic framework for the analysis of this process is defined. Practical tools that support the bootstrapping process are developed and the efficiency of the process is analysed from both a machine learning and a human factors perspective. We find that even linguistically untrained users can use the system to create electronic pronunciation dictionaries accurately, in a fraction of the time the traditional approach requires. We create new dictionaries in a number of languages (isiZulu, Afrikaans and Sepedi) and demonstrate the utility of these dictionaries by incorporating them in speech technology systems. / Thesis (PhD (Electronic Engineering))--University of Pretoria, 2006. / Electrical, Electronic and Computer Engineering / unrestricted

We present a novel method for building 3D models from a user sketch. Given a 2D sketch as input, the approach aligns and deforms a chosen 3D template model to match the sketch. This is guided by a set of user-specified correspondences and an algorithm that deforms the 3D model to match the sketched profile. Our primary contribution is related to fitting the 3D deformable geometry to the 2D user sketch. We demonstrate our technique on several examples. / Science, Faculty of / Computer Science, Department of / Graduate

Cheap electronic storage and Internet bandwidth has increased the amount of online data. Large quantities of metadata are created to manage this wealth of information. Methods to organize and structure metadata has led to the development of ontologies - data that is organized to describe the relation between elements. The creation of large ontologies has brought forth the need for ontology management strategies. Ontology alignment and merging techniques are standard operations for ontology management. Accurate ontology alignment methods are typically semi-automatic, meaning they require periodic user input. This becomes infeasible on large ontologies and the accuracy and efficiency drops significantly when these algorithms are forced to align without human interaction. Bioinformatics, for example, has seen the influx of large ontologies, such as signal pathway sets with thousands of elements or protein-protein interaction (PPI) databases with hundreds of thousands of elements. This drives the need for a reliable method of large-scale ontology alignment. Many bioinformatics ontologies contain references to domain ontologies - manually curated ontologies describing additional, general information about the terms in the ontologies. For example, more than 2/3 of proteins in PPI data sets contain at least one annotation to the domain ontology the Gene Ontology. We use the domain ontology references as features to compute similarity between elements. However, there are few efficient ways to compute similarity from structured features. We present a novel, automatic method for aligning ontologies based on such domain ontology features. Specifically, we use simulated annealing to reduce the complexity of the domain ontologys structure by finding approximate relevant clusters of elements. An intermediate step performs hierarchical clustering based on the similarity between elements of the ontology. Then the mapping between clusters across aligning ontologies is built. The final step builds an alignment between matched clusters. To evaluate our methods, we perform an alignment between Human (Homo Sapiens) and Yeast (Saccharomyces cerevisiae) signal pathways provided by the Reactome database. The results were compared against reliable homology studies of proteins. The final mapping produces alignments that are significantly more accurate than the traditional ontology alignment methods, without any human involvement. / Science, Faculty of / Computer Science, Department of / Graduate

This research study focuses on examining the views of patients, their families, and staff member providers of the Centre for Healthy Active Living a clinical obesity management program for children and youth at the Children’s Hospital of Eastern Ontario, in Ottawa, Canada. Qualitative methodology was used and content analysis was conducted with data obtained from family questionnaires and a provider focus group. Analysis of obtained data was conducted to determine alignment of views between patient, family, and provider views, and the formal goals of the program. Emerging themes from the data indicated that patients and families place higher value on the formal goals of “improve quality of life; improve eating behaviours; improving fitness, increasing activity levels; and empower/strengthen families.” (Children’s Hospital of Eastern Ontario, 2012). Specific recommendations with regards to each component of these goals were provided. Instrumental, procedural, systemic, and conceptual recommendations of program components were also provided.

Bilingual terminological resources play a pivotal role in human and machine translation of technical text. Owing to the immense volume of newly produced terminology in the biomedical domain, existing resources suffer from low coverage and they are only available for a limited number of languages. The need for term alignment methods that accurately identify translations of terms, emerges. In this work, we focus on bilingual terminology induction from freely available comparable corpora, i.e. thematically related documents in two or more languages. We investigate different sources of information that determine translation equivalence, including: (a) the internal structure of terms (compositional clue), (b) the surrounding lexical context (contextual clue) and (c) the topic distribution of terms (topical clue). We present four novel compositional alignment methods and we introduce several extensions over existing compositional, context-based and topic-based approaches. Furthermore, we combine the three translation clues in a single term alignment model and we show substantial improvements over the individual translation signals when considered in isolation. We examine the performance of the proposed term alignment methods on closely related (English-French, English-Spanish) language pairs, on a more distant, low-resource language pair (English-Greek) and on an unrelated (English-Japanese) language pair. As an application, we integrate automatically compiled bilingual terminologies with Statistical Machine Translation systems to more accurately translate unknown terms. Results show that an up-to-date bilingual dictionary of terms improves the translation performance of SMT.

This thesis sets out to investigate the field of public service interpreting in southern Spain, with a particular emphasis on the position of volunteer interpreters working at two different healthcare institutions. It looks at the power relationships that develop between agents that hold different degrees of control and autonomy, especially in a context where individuals hold different forms and volume of capital in each encounter. Drawing on Bourdieu’s Theory of Practice, the study offers an in-depth examination of a group of volunteer interpreters as legitimate agents of the wider field of public service interpreting and the sub-field of healthcare interpreting, while looking at their impact on the structures and ethics of the larger field. This is the first project to employ Bourdieu’s theory in a sustained case study of a healthcare context where volunteer interpreters operate as legitimised institutional agents. One of the peculiarities of the two settings under examination is that volunteer interpreters seem to have acquired a high degree of institutionalisation, which provides them with a large volume of symbolic capital and allows them to take part in the field as legitimate members of the healthcare team, often occupying similar positions to those adopted by doctors at the top end of the field hierarchy. The study adopts an ethnographic approach based on a triangulation of data: participant observation of volunteer interpreters, audio-recorded interpreter-mediated interaction and focus-group interviews with volunteer interpreters. The primary data that informs the thesis consists of four focus groups carried out with volunteer interpreters in two different Spanish hospitals. The additional use of participant observations and audio-recordings make it possible to examine not only interpreters’ perceptions but also actual behaviour in authentic encounters, and to compare interpreters’ perception of their positioning with the actual positions they often occupy in the field.

The thesis Application of Balanced Scorecard is focused on creation of strategic management system BSC for company Bayer, s. r. o. and its division Bayer MaterialScience. In theoretical part the system BSC and its four perspectives are introduced. In practical part the company Bayer, s. r. o., divisions, strategy and vision are introduced. There is a SWOT analysis of Bayer, s. r. o. and Bayer MaterialScience created and then the system Balanced Scorecard is applied. The thesis also includes system alignment of company and division. The system Balanced Scorecard is described by strategic maps that are included in annex.

With the current drive of both academia and practice to ensure that there is an appropriate degree of collaboration to ensure mutual benefit, any study undertaken that has the application of theory as one of its objectives, should therefore be able to make a contribution to both areas. From a strategic ICT planning perspective the ability to apply a planning process in a cohesive and coordinated manner throughout diversified or complex organisations has been one of the challenges that faced both academics and practitioners. This becomes especially relevant in view of the drive to ensure alignment between business and ICT so as to ensure that the potential utility of ICT can be leveraged and sustained to contribute towards the competitive advantage of the organisation. This situation is also relevant to the South African Department of Defence as a particular kind of diversified organisation. With this in mind the initiative was recognised to establish an appropriate strategic ICT planning process for the department and to apply it to the point where a corporate strategic ICT master plan could be institutionalised. A progressive and longitudinal action research process resulted in not only the development of an appropriate process of definition and alignment, but also on the ability to apply and manage the strategic ICT planning function with full collaboration of all role players and stakeholders. The conclusion from this research was that it is indeed possible to develop and apply an appropriate strategic ICT planning process that recognises the strategic – corporate and business unit level – environment of a diversified organisation, but that it is largely dependant upon the ability to ensure collaboration and participation within a clear and distinct construct of functions and responsibilities. Without appropriate management arrangements and mechanisms, it was found to be very difficult indeed due to issues relating to the semi-autonomous nature of the respective business units in relation to corporate management. / Thesis (PhD(IT))--University of Pretoria, 2008. / Informatics / unrestricted

Wireless communication systems utilize the wireless medium to perform over-the-air (OTA) data transfer. There are many factors that can impact the quality of wireless communications, such as medium imperfection, interfering environment, mismatch of transceivers, etc. To mitigate these problems and improve the quality of service (QoS), this research study is conducted on three important topics including synchronization techniques, impairment estimation theory and techniques, and interference alignment techniques. In this thesis, it firstly present a dual link algorithm to align and manage the interference of multiple-input and multiple-output (MIMO) networks. A field-programmable gate array (FPGA) prototype is designed for software defined radio (SDR) platforms. As one of the key components, a hardware efficient architecture is proposed for the implementation of singular value decomposition (SVD). Secondly, it proposes a maximum-likelihood (ML) based synchronization approach for carrier frequency synchronization for MIMO systems. The algorithm is also implemented on FPGA for real-time performance evaluation. Finally, as an exemplary study of machine learning techniques for wireless communications, a neural network (NN) based estimator is proposed to perform coarse frequency offset estimations for MIMO systems. The proposed NN based estimator can accommodate various channel models and the results show promising performance in terms of accuracy and estimation range. In summary, this thesis provides a comprehensive study on interference alignment, carrier synchronization, and impairment estimation using different approaches. Efficient hardware implementations for the key algorithms are also presented.

For more than seventy years, semiconductor devices have functioned as the cornerstone for technological advancement, and as the defining transition into the information age. The III-Nitride family of semiconductors, in particular, underwent an impressive maturation over the past thirty years, which allowed for efficient light- emitting devices, photo-detectors, and power electronic devices. As researchers try to push the limits of semiconductor devices, and in particular, as they aim to design ultraviolet light emitters and high threshold power devices, the search for new materials with high band gaps, high breakdown voltages, unique optical properties, and variable lattice parameters is becoming a priority. Two interesting candidates that can help in achieving the aforementioned goals are the wurtzite BAlN and BGaN alloy systems, which are currently understudied due to difficulties associated with their growth in epitaxial settings. In our research, we will investigate the band alignment between BAlN and BGaN alloys, and other wurtzite III-Nitride semiconductors from first principle simulations. Through an understanding of band alignment types and a quantification of the band offset values, researchers will be able to foresee the applicability of a particular interface. As an example, a type-I band alignment with a high conduction band offset and a low valence band offset is a potential electron blocking layer to be implemented in standard LED designs. This first principle investigation will be aided by simulations using Density Functional Theory (DFT) as implemented in the Vienna Ab Initio Simulation Package (VASP) environment. In addition, we will detail an experiment from the literature that uses X- ray Photoelectron Spectroscopy on multiple samples to infer and quantify the band alignment between different materials of interest to us. We aim in this study to anticipate the band alignment in interfaces involving materials at the cutting edge of research. Our hope is to set a theoretical ground for future experimental studies on this same matter in parallel to the current efforts to improve the quality and stability of wurtzite BAlN and BGaN alloy crystals.