A data warehouse is a single data store, where data from multiple data sources is integrated for online business analytical processing (OLAP) of an entire organisation. The rationale being single and integrated is to ensure a consistent view of the organisational business performance independent from different angels of business perspectives. Due to its wide coverage of subjects, data warehouse design is a highly complex, lengthy and error-prone process. Furthermore, the business analytical tasks change over time, which results in changes in the requirements for the OLAP systems. Thus, data warehouse and OLAP systems are rather dynamic and the design process is continuous. In this thesis, we propose a method that is integrated, formal and application-tailored to overcome the complexity problem, deal with the system dynamics, improve the quality of the system and the chance of success. Our method comprises three important parts: the general ASMs method with types, the application tailored design framework for data warehouse and OLAP, and the schema integration method with a set of provably correct refinement rules. By using the ASM method, we are able to model both data and operations in a uniform conceptual framework, which enables us to design an integrated approach for data warehouse and OLAP design. The freedom given by the ASM method allows us to model the system at an abstract level that is easy to understand for both users and designers. More specifically, the language allows us to use the terms from the user domain not biased by the terms used in computer systems. The pseudo-code like transition rules, which gives the simplest form of operational semantics in ASMs, give the closeness to programming languages for designers to understand. Furthermore, these rules are rooted in mathematics to assist in improving the quality of the system design. By extending the ASMs with types, the modelling language is tailored for data warehouse with the terms that are well developed for data-intensive applications, which makes it easy to model the schema evolution as refinements in the dynamic data warehouse design. By providing the application-tailored design framework, we break down the design complexity by business processes (also called subjects in data warehousing) and design concerns. By designing the data warehouse by subjects, our method resembles Kimball's "bottom-up" approach. However, with the schema integration method, our method resolves the stovepipe issue of the approach. By building up a data warehouse iteratively in an integrated framework, our method not only results in an integrated data warehouse, but also resolves the issues of complexity and delayed ROI (Return On Investment) in Inmon's "top-down" approach. By dealing with the user change requests in the same way as new subjects, and modelling data and operations explicitly in a three-tier architecture, namely the data sources, the data warehouse and the OLAP (online Analytical Processing), our method facilitates dynamic design with system integrity. By introducing a notion of refinement specific to schema evolution, namely schema refinement, for capturing the notion of schema dominance in schema integration, we are able to build a set of correctness-proven refinement rules. By providing the set of refinement rules, we simplify the designers's work in correctness design verification. Nevertheless, we do not aim for a complete set due to the fact that there are many different ways for schema integration, and neither a prescribed way of integration to allow designer favored design. Furthermore, given its °exibility in the process, our method can be extended for new emerging design issues easily.

This thesis relates to research and development in the field of 3D mesh data for computer graphics. A review of existing storage and manipulation techniques for mesh data is given followed by a framework for mesh editing. The proposed framework combines complex mesh editing techniques, automatic level of detail generation and mesh compression for storage. These methods work coherently due to the underlying data structure. The problem of storing and manipulating data for 3D models is a highly researched field. Models are usually represented by sparse mesh data which consists of vertex position information, the connectivity information to generate faces from those vertices, surface normal data and texture coordinate information. This sparse data is sent to the graphics hardware for rendering but must be manipulated on the CPU. The proposed framework is based upon geometry images and is designed to store and manipulate the mesh data entirely on the graphics hardware. By utilizing the highly parallel nature of current graphics hardware and new hardware features, new levels of interactivity with large meshes can be gained. Automatic level of detail rendering can be used to allow models upwards of 2 million polygons to be manipulated in real time while viewing a lower level of detail. Through the use of pixels shaders the high detail is preserved in the surface normals while geometric detail is reduced. A compression scheme is then introduced which utilizes the regular structure of the geometry image to compress the floating point data. A number of existing compression schemes are compared as well as custom bit packing. This is a TIF funded project which is partnered with Unlimited Realities, a Palmerston North software development company. The project was to design a system to create, manipulate and store 3D meshes in a compressed and easy to manipulate manner. The goal is to create the underlying technologies to allow for a 3D modelling system to become integrated into the Umajin engine, not to create a user interface/stand alone modelling program. The Umajin engine is a 3D engine created by Unlimited Realities which has a strong focus on multimedia. More information on the Umajin engine can be found at www.umajin.com. In this project we propose a method which gives the user the ability to model with the high level of detail found in packages aimed at creating offline renders but create models which are designed for real time rendering.

Online banking systems have become more common and widely used in daily life, bringing huge changes in modern banking transaction activities and giving us a greater opportunity to access the banking system anytime and anywhere. At the same time, however, one of the key challenges that still remain is to fully resolve the security concerns associated with the online banking system. Many clients feel that online banking is not secure enough, and to increase its security levels, many banks simply add more identity-checking steps or put on more security measures to some extent to give users the impression of a secure online banking system. However, this is easier to be said than done, because we believe that more identity-checking steps could compromise the usability of the online banking system, which is an inevitable feature in design of usable and useful online banking systems. Banks can simply enhance their security level with more sophisticated technologies, but this does not seem to guarantee the online banking system is in line with its key usability concern. Therefore, the research question raised in this thesis is to establish the relationships between usability, security and trustworthiness in the online banking system. To demonstrate these relationships, three experiments were carried out using the simulation of an online banking logon procedure to provide a similar online banking experience. Post questionnaires were used to measure the three concepts, i.e. usability, security and trustworthiness. The resulting analyses revealed that simply adding more identity-checking steps in the online banking system did not improve the customers? perceived security and trustworthiness, nor the biometric security technique (i.e., fingerprints) did enhance the subjective ratings on the perceived security and trustworthiness. This showed that the systems designer needs to be aware that the customer?s perception of the online banking system is not the same as that conceived from a technical standpoint.

The use of computers has become part of every day life. The high prevalence of computer use may lead employers to assume university graduates will have good computing skills. Such assumptions may be the reason that employers use broad terms to advertise the computing tasks required for graduate-level positions. This thesis investigates how well the expectations of employers match the perceptions of near-graduates about their computer skills. Four graduate-level positions were identified from advertisements placed in order to recruit graduates. The employers who placed these advertisements were surveyed by interview and questionnaire. Twenty-one students about to graduate from a university commerce programme were also interviewed and surveyed. It was found that the wording of the advertisements did not satisfactorily portray the requirements and intentions of the employers. It was also found that skills the near-graduates perceived they possessed frequently did not meet the expectations of employers. Results also show that the near-graduates did not fully understand which computing skills would be expected in the workplace. This study highlights implications for three groups: employers, graduates and educators.

This thesis seeks to address current problems encountered when training classifiers within the framework of cascades of boosted ensembles (CoBE). At present, a signifi- cant challenge facing this framework are inordinate classifier training runtimes. In some cases, it can take days or weeks (Viola and Jones, 2004; Verschae et al., 2008) to train a classifier. The protracted training runtimes are an obstacle to the wider use of this framework (Brubaker et al., 2006). They also hinder the process of producing effective object detection applications and make the testing of new theories and algorithms, as well as verifications of others research, a considerable challenge (McCane and Novins, 2003). An additional shortcoming of the CoBE framework is its limited ability to train clas- sifiers incrementally. Presently, the most reliable method of integrating new dataset in- formation into an existing classifier, is to re-train a classifier from beginning using the combined new and old datasets. This process is inefficient. It lacks scalability and dis- cards valuable information learned in previous training. To deal with these challenges, this thesis extends on the research by Barczak et al. (2008), and presents alternative CoBE frameworks for training classifiers. The alterna- tive frameworks reduce training runtimes by an order of magnitude over common CoBE frameworks and introduce additional tractability to the process. They achieve this, while preserving the generalization ability of their classifiers. This research also introduces a new framework for incrementally training CoBE clas- sifiers and shows how this can be done without re-training classifiers from beginning. However, the incremental framework for CoBEs has some limitations. Although it is able to improve the positive detection rates of existing classifiers, currently it is unable to lower their false detection rates.

This thesis seeks to address current problems encountered when training classifiers within the framework of cascades of boosted ensembles (CoBE). At present, a signifi- cant challenge facing this framework are inordinate classifier training runtimes. In some cases, it can take days or weeks (Viola and Jones, 2004; Verschae et al., 2008) to train a classifier. The protracted training runtimes are an obstacle to the wider use of this framework (Brubaker et al., 2006). They also hinder the process of producing effective object detection applications and make the testing of new theories and algorithms, as well as verifications of others research, a considerable challenge (McCane and Novins, 2003). An additional shortcoming of the CoBE framework is its limited ability to train clas- sifiers incrementally. Presently, the most reliable method of integrating new dataset in- formation into an existing classifier, is to re-train a classifier from beginning using the combined new and old datasets. This process is inefficient. It lacks scalability and dis- cards valuable information learned in previous training. To deal with these challenges, this thesis extends on the research by Barczak et al. (2008), and presents alternative CoBE frameworks for training classifiers. The alterna- tive frameworks reduce training runtimes by an order of magnitude over common CoBE frameworks and introduce additional tractability to the process. They achieve this, while preserving the generalization ability of their classifiers. This research also introduces a new framework for incrementally training CoBE clas- sifiers and shows how this can be done without re-training classifiers from beginning. However, the incremental framework for CoBEs has some limitations. Although it is able to improve the positive detection rates of existing classifiers, currently it is unable to lower their false detection rates.

This thesis seeks to address current problems encountered when training classifiers within the framework of cascades of boosted ensembles (CoBE). At present, a signifi- cant challenge facing this framework are inordinate classifier training runtimes. In some cases, it can take days or weeks (Viola and Jones, 2004; Verschae et al., 2008) to train a classifier. The protracted training runtimes are an obstacle to the wider use of this framework (Brubaker et al., 2006). They also hinder the process of producing effective object detection applications and make the testing of new theories and algorithms, as well as verifications of others research, a considerable challenge (McCane and Novins, 2003). An additional shortcoming of the CoBE framework is its limited ability to train clas- sifiers incrementally. Presently, the most reliable method of integrating new dataset in- formation into an existing classifier, is to re-train a classifier from beginning using the combined new and old datasets. This process is inefficient. It lacks scalability and dis- cards valuable information learned in previous training. To deal with these challenges, this thesis extends on the research by Barczak et al. (2008), and presents alternative CoBE frameworks for training classifiers. The alterna- tive frameworks reduce training runtimes by an order of magnitude over common CoBE frameworks and introduce additional tractability to the process. They achieve this, while preserving the generalization ability of their classifiers. This research also introduces a new framework for incrementally training CoBE clas- sifiers and shows how this can be done without re-training classifiers from beginning. However, the incremental framework for CoBEs has some limitations. Although it is able to improve the positive detection rates of existing classifiers, currently it is unable to lower their false detection rates.

The concept of Multi-User Virtual Environments (MUVEs) has opened new avenues in the educational spectrum. Despite its popularity as an educational environment tool, the successful implementation of a virtual classroom is heavily reliant on the educator. This research focuses on the perceptions of educators regarding the acceptance of the MUVE as an educational tool. The Technology Acceptance Model (TAM) was used to identify and evaluate the potential benefits of the MUVE in the domain of education. The qualitative approach was considered to be the suitable approach for this study. Semi-structured interviews were conducted with 22 educators; these interviews included the demonstration of a virtual class located in the Second Life Island known as Jokaydia. The collected data was transcribed using NVivo software, and analysed using constant comparison analysis. The transcribed interviews were provided to another researcher in order to obtain an independent analysis; this created the basis for triangulation of participants’ perceptions. A summary of this analysis was then sent to all participants to confirm its credibility. The conclusions of the study suggest that the combination of MUVEs’ features and strengths will eventually influence the educators to accept the MUVE as an educational tool, although several areas of concern are identified. Future growth in the educational uses of MUVEs is examined, the implications and limitations of the study are discussed, and ideas for future research are elaborated on. Keywords: MUVE, Second Life, education, TAM, ease of use, subjective norm, enjoyment, facilities, compatibility, security and trust, collaboration, awareness, media richness, discovery learning, situated learning, role playing, controlled environment, immersiveness.

This thesis seeks to address current problems encountered when training classifiers within the framework of cascades of boosted ensembles (CoBE). At present, a signifi- cant challenge facing this framework are inordinate classifier training runtimes. In some cases, it can take days or weeks (Viola and Jones, 2004; Verschae et al., 2008) to train a classifier. The protracted training runtimes are an obstacle to the wider use of this framework (Brubaker et al., 2006). They also hinder the process of producing effective object detection applications and make the testing of new theories and algorithms, as well as verifications of others research, a considerable challenge (McCane and Novins, 2003). An additional shortcoming of the CoBE framework is its limited ability to train clas- sifiers incrementally. Presently, the most reliable method of integrating new dataset in- formation into an existing classifier, is to re-train a classifier from beginning using the combined new and old datasets. This process is inefficient. It lacks scalability and dis- cards valuable information learned in previous training. To deal with these challenges, this thesis extends on the research by Barczak et al. (2008), and presents alternative CoBE frameworks for training classifiers. The alterna- tive frameworks reduce training runtimes by an order of magnitude over common CoBE frameworks and introduce additional tractability to the process. They achieve this, while preserving the generalization ability of their classifiers. This research also introduces a new framework for incrementally training CoBE clas- sifiers and shows how this can be done without re-training classifiers from beginning. However, the incremental framework for CoBEs has some limitations. Although it is able to improve the positive detection rates of existing classifiers, currently it is unable to lower their false detection rates.

This thesis seeks to address current problems encountered when training classifiers within the framework of cascades of boosted ensembles (CoBE). At present, a signifi- cant challenge facing this framework are inordinate classifier training runtimes. In some cases, it can take days or weeks (Viola and Jones, 2004; Verschae et al., 2008) to train a classifier. The protracted training runtimes are an obstacle to the wider use of this framework (Brubaker et al., 2006). They also hinder the process of producing effective object detection applications and make the testing of new theories and algorithms, as well as verifications of others research, a considerable challenge (McCane and Novins, 2003). An additional shortcoming of the CoBE framework is its limited ability to train clas- sifiers incrementally. Presently, the most reliable method of integrating new dataset in- formation into an existing classifier, is to re-train a classifier from beginning using the combined new and old datasets. This process is inefficient. It lacks scalability and dis- cards valuable information learned in previous training. To deal with these challenges, this thesis extends on the research by Barczak et al. (2008), and presents alternative CoBE frameworks for training classifiers. The alterna- tive frameworks reduce training runtimes by an order of magnitude over common CoBE frameworks and introduce additional tractability to the process. They achieve this, while preserving the generalization ability of their classifiers. This research also introduces a new framework for incrementally training CoBE clas- sifiers and shows how this can be done without re-training classifiers from beginning. However, the incremental framework for CoBEs has some limitations. Although it is able to improve the positive detection rates of existing classifiers, currently it is unable to lower their false detection rates.