Abstracts

For a digital copy of the proceedings, refer to the Papers page.

Full Papers

Anu Vaidyanathan, Mark Billinghurst and Harsha Sirisena: Characterizing and Visualizing Mobile Networks

This paper seeks to characterize data obtained from calls made on a mobile phone network, from one of New Zealand’s leading telecommunications providers. By understanding the characteristics of the data, we provide insight into the metrics, which are important to keep in sight for modelling such data for use in our performance visualization tool, Caviar. The primary characteristics that are of interest to us include the user’s mobility across the board, while calls are being placed on the network. In understanding the co-relation of the users’ activity as opposed to mobility, gathering insight into the frequency, duration and median cell-site changes, we provide the foundation for data-characteristics, which can be used to visualize performance of the mobile network, besides being used in traffic monitoring, resource-usage planning, social networks, proactive error-correction etc. in order for the telecommunications service provider to maintain their clientele and provide other services.

Carl Schultz, Robert Amor and Hans W. Guesgen: A Framework for Applying Qualitative Spatial and Temporal Reasoning

Many application domains make extensive use of spatial and temporal information. However, the numerical approaches employed by most software tools have limitations, particularly when information is vague or incomplete. To address this, alternative qualitative spatial and temporal reasoning (QSTR) methods have been developed, yet few applications have made significant use of these techniques. In response to this we are developing a framework that will support the application of QSTR by allowing software developers to create custom qualitative modelling systems. In this paper we compare our framework to more standard toolbox approaches for QSTR application support. We present fundamental principles of qualitative modelling, and demonstrate, using an architectural lighting example, how these principles provide a basis for creating qualitative modelling systems that incorporate domain knowledge.

Mayank Keshariya and Ray Hunt: An Overview of Proposed Layered-Model and its Architecture for Extending Policy-based Management Framework to Provide Always Best Connected services

Trusted Network Connect (TNC) is a network access control mechanism that takes the security state of access requesting endpoints into account. This mechanism is currently limited to local area networks and VPN connections. However, TNC can further be useful in other scenarios, such as security sensitive web-based applications. An architectural approach for TNC in this environment based on authentication standards is presented in this paper.

Felix Manke and Burkhard C. Wuensche: Fast Three-Dimensional Texture Synthesis

Three-dimensional (3D) textures are popular for texturing solid objects. They usually achieve superior results over conventional two-dimensional (2D) texture mapping techniques. However, whereas 2D textures can be obtained using for example a photograph, capturing or generating solid textures is not straightforward. Procedural techniques can replicate some types of materials, but are usually difficult to control. An alternative is exemplar-based texture synthesis where 3D textures are generated from a few 2D pictures. Unfortunately, the synthesis from exemplars is very challenging and usually requires very long computation times. In this paper we present a new algorithm for fast solid texture synthesis from 2D input exemplars. Our method extends an existing real-time synthesis approach which has so far been limited to 2D texture synthesis only. The modifications we made allow a hardware-accelerated synthesis utilising the computational capabilities of today’s graphics cards. To our knowledge, our method is the first solid texture synthesis approach that was specifically designed for an execution directly on graphics hardware.

Gian Perrone and David Streader: Automatic Parallelisation of Web Applications

Small web applications have a tendency to get bigger. Yet despite the current popularity of web applications, little has been done to help programmers to leverage the performance and scalability benefits that can result from the introduction of parallelism into a program. Accordingly, we present a technique for the automatic parallelisation of whole web applications, including persistent data storage mechanisms. We detail our prototype implementation of this technique (“Ceth”) and finally, we establish the soundness of the process by which we extract coarse-grained parallelism from programs.

Mohammad Daradkeh, Clare Churcher and Alan McKinnon: Interactive Visualization Techniques for Exploring Model Sensitivity

Visualization can play an important role in uncertainty and sensitivity analysis. It provides a means for graphically exploring the relationships between the output and the inputs of a model and to determine how “sensitive” a model is to changes in the values of the input. The literature shows how sensitivity analysis can contribute to improved decision making, but little can be found about the advantages of exploring model sensitivity visually to aid the decision maker. We aim to develop new interactive visualization techniques to assist people who are using models for decision making but who need to explore the often complex relationships between the values of model variables and the model output.

Sergio Hernandez: On Line Bayesian Tracking and Detection of Multiple Objects

Sequential Monte Carlo (SMC) methods such as particle filters have been used in tracking problems for moving from an intractable distribution to a density that is closer to the actual posterior distribution. These methods make use of stochastic simulations that can approximate non-linear and non-Gaussian posterior distributions via importance sampling. Since the standard SMC methods only allow to track a single target, it becomes necessary find representations of the multiple object posterior density. The filtering distribution must take into account measurement-to-track data association and the probability of objects appearing and disappearing in the field of view. The problem can be even more challenging when considering misdetections and clutter, so methods based on random sets and the point process theory have been proposed as suitable representations for the intractable distribution.

Scott Raynel and Murray Pearson: An Overview of Link-Level Measurement Techniques for Wide-Area Wireless Networks

By building wireless link-level measurement tools we hope to improve the design, deployment and management of wide-area wireless community networks. This paper identifies existing link-level measurement techniques and discusses the advantages and disadvantages of each in the context of measuring and monitoring such networks. Finally, we make a case for the need for more sophisticated techniques and tools which will assist both day-to-day network operations as well as wireless network research.

Shaoqun Wu: How to Express Your Feelings (With a Little Help from Google)

This paper describes an attempt to capitalize on the vast amount of human-generated text readily available on the Web to help language learners express their own feelings. We avoid errors, idiosyncrasies, and other dross by employing various language and grammar filters. We sort words and phrases by frequency of use to ensure that only very common usage is included. The work is based on a huge collection of n-grams published by Google, along with their frequency on the web. We describe a comprehensive query tool that allows language learners and teachers alike to locate a particular word, its associated phrases, clause patterns, synonyms and antonyms. In addition, we have enabled sample sentences containing these patterns to be retrieved from the Web and presented to the user. Finally, five language activities have been designed to help learners master important vocabulary and expressions. The work is at a preliminary stage and no user tests have yet been performed.

Sebastian Hoehna: Evaluation of Proposal Distributions on Clock-Constrained Trees in Bayesian Phylogenetic Inference

Bayesian Markov chain Monte Carlo (MCMC) has become one of the principle methods of performing phylogenetic inference. Implementing the Markov chain Monte Carlo algorithm requires the definition of a proposal distribution which defines a transition kernel over the state space. The precise form of this transition kernel has a large impact on the computational efficiency of the algorithm. In this paper we investigate the efficiency of a number of different proposal distributions for clock-constrained phylogenetic trees (i.e. constrained by a strict or relaxed molecular clock). Clock-constrained trees have become increasingly important in phylogenetic inference, especially in the context of divergence time estimation and their constraints require substantially different proposal algorithms to unrooted phylogenetic trees. We investigated the efficiency of seven proposal moves on clock-constraint trees first on a small data set and then on six additional data sets. In contrast to the results for the case of MCMC on unconstrained phylogenetic trees we found that subtree swapping moves perform better than subtree prune and regraft algorithms and moderate proposals dominate bold proposals. However, the results varied with the data set we analyzed and the intermediate subtree swap proposal distribution which we introduce in this paper was the only one with a continuous high level of efficiency.

Jesse Read: A Pruned Problem Transformation Method for Multi-label Classification

Multi-label classification has gained significant interest in recent years, paralleled by the increasing use of manual multi-labelling, often known as applying “tags” to documents. Well known examples include Flickr, YouTube, CiteULike and Google Bookmarks. This paper focuses on Problem Transformation (PT) as an approach to multi-label classification and details these methods as well as their respective advantages and disadvantages. A Pruned Problem Transformation method (PPT) is presented, along with several extensions, designed to overcome such disadvantages. This new method is empirically compared with existing methods, both in terms of accuracy and training time, and the results are encouraging.

Jason Alexander and Andy Cockburn: Characterising Electronic Document Use, Reuse, Coverage and Multi-Document Interaction

What documents do you use? How much of your document are others likely to read? How much time do you spend using documents? Desktop electronic document manipulation is one of the most common activities performed by computer users, yet there remains little empirical research into how documents are used in common document navigation systems. This paper presents a 14 participant, 120 day study that logged user actions in Microsoft Word and Adobe Reader, with the aim of characterising document use. The study found that Microsoft Word documents are likely to be significantly shorter but have longer periods of interaction compared to Adobe Reader documents. Word documents averaged 6 pages in length and Reader documents 38 pages. Documents that were ten pages or less made up 80% of those that were opened. Approximately half of the documents viewed were reopenings of ones previously used, however history mechanisms were poorly utilised. Document coverage in Microsoft Word was approximated by a normal distribution, while Reader document coverage decreased in a linear fashion, the further one moved toward the end. The time spent with multiple documents open decreased exponentially as the number of documents open increased. We briefly discuss the implications these findings have for the design of document navigation systems.

Daniel McEnnis: Towards a Music Recommendation Infrastructure

Contrary to opinions widely voiced in the popular media, music is not a universal language. Music exists in all cultures, but what it means and how it is interpreted differs on many levels. This paper seeks to break away from data-driven analysis that makes implicit assumptions about musical properties common among all cultures and replace it with a framework built on social sciences research. The Relational Analysis Toolkit (RAT) provides structure and support for building musical recommendation systems that are not ethno-centric and integrates differences in musical taste and understanding across every level of analysis: intra-personal, inter-personal, geographical and cultural levels.

Rami Mounla: QoS-Aware Web Service Composition

With the increase of Web service popularity, due to its well established secure protocols, service providers are competing in the market with Web services that provide overlapping, if not similar, functionalities. By improving the Web services quality, they become more attractive to customers. When considering complex composite Web services, the selection of the most appropriate services becomes a difficult task. Previous research suggests the use of integer programming to solve the selection problem. In this paper a hybrid approach using integer programming and case-based reasoning will be presented. The compositions will be globally optimized using integer programming and store in a case-based reasoning repository for future use. A set of experiments will show that not only does the system perform better than traditional approaches, but that it also has the ability to “learn” quality patterns and solve the composition problem based on those patterns.

Guy K. Kloss, Napoleon Reyes and Ken Hawick: Gaining Colour Stability in Live Image Capturing

Digital colour cameras are dramatically falling in price, making them affordable for ubiquitous appliances in many applications. An attempt to use colour information reveals a significant problem that usually escapes our awareness. Due to the adaptive nature of the human visual system in most cases we do not recognise most changes in illumination characteristics, a camera however will measure scenes under changing illumination differently. Attempts to deduce object colour from the images will need to cope with the influence of the illumination and the camera’s characteristics. Furthermore, a large variety of colour spaces are available to describe colour. Differences between them and their fitness to quantify colour are discussed. This paper tries to establish a basic understanding of the intricacies behind the processes involved in capturing images and recognising colour - from light as a stimulus to the colour sensed values in cameras. The goal is to outline a novel approach fusing common industrial best practices with dynamic adaptation capabilities needed for robustly measuring colour using cameras in real-time. First positive results towards improving colour based reasoning on adaptable colour spaces are stated as an outlook for further development directions.

Olena Medelyan and David Milne: Augmenting Domain-Specific Thesauri with Knowledge from Wikipedia

We propose a new method for extending a domain-specific thesaurus with valuable information from Wikipedia. The main obstacle is to disambiguate thesaurus concepts to correct Wikipedia articles. Given the concept name, we first identify candidate mappings by analyzing article titles, their redirects and disambiguation pages. Then, for each candidate, we compute a link-based similarity score to all mappings of context terms related to this concept. The article with the highest score is than used to augment the thesaurus concept. It is the source for the extended gloss, explaining the concept’s meaning, synonymous expressions that can be used as additional non-descriptors in the thesaurus, translations of the concept into other languages, and new domain-relevant concepts.

Norsaremah Salleh: A Systematic Review of Pair Programming Research - Initial Results

This paper presents the initial results of our systematic review of pair programming studies. As an alternative to performing a literature review, we conducted a systematic review in order to reveal the answers to our research questions pertaining to the issues of pair programming in an educational context. Our focus is to identify factors affecting the effectiveness of students who pair programmed, particularly related to the psychosocial factors such as compatibility, personality and gender issues. Our systematic review follows the phases described in the procedures for performing systematic reviews outlined in [18]. We have included 66 studies for the synthesis of evidence, of which 10 studies were found relevant to answer the first research question. The findings showed that personality type is the most common factor investigated. However, the results of those studies were somewhat inconclusive.

Parma Nand: On the Use of Salience Weights in Anaphora Resolution

This paper presents the results of research undertaken on anaphora resolution achieved by partially-developed algorithm based solely on salience weights. The research, which is part of a PhD, is focused on resolving anaphors focused specifically in the genre of in short newspaper type articles because it forms part of wider research aimed at building a system for visualization of online newspaper articles. The paper discusses the characteristics of such a corpus and describes an initial implementation of an algorithm to resolve the anaphors using knowledge-poor approach which is completely based on salience scores. Salience scores have been widely used by various researchers in conjunction with other techniques but it has not been exclusively used in any algorithm. In this paper I present results from an algorithm which uses salience weights exclusively including an analysis of the rules imposed by the magnitudes of the various salience weights. The algorithm described is currently using no integrated knowledge including morphology of nouns which is commonly integrated even in algorithms classified as knowledge-poor. It works on semi-parsed text and currently only uses syntax and surface level semantics of the language to resolve primarily pronominal pronouns. The results presented use two articles form the local newspaper (New Zealand Herald). The input data is also tested on two publicly available anaphora resolution algorithms and the results compared to the algorithms described in this paper. Finally it discusses the challenges in developing light weight algorithms and scope for further improvements including integration of knowledge to resolve associative anaphora.

Anna Huang: Similarity Measures for Text Document Clustering

Clustering is a useful technique that organizes a large quantity of unordered text documents into a small number of meaningful and coherent clusters, thereby providing a basis for intuitive and informative navigation and browsing mechanisms. Partitional clustering algorithms have been recognized to be more suitable as opposed to the hierarchical clustering schemes for processing large datasets. A wide variety of distance functions and similarity measures have been used for clustering, such as squared Euclidean distance, cosine similarity, and relative entropy. In this paper, we compare and analyze the effectiveness of these measures in partitional clustering for text document datasets. Our experiments utilize the standard K-means algorithm and we report results on seven text document datasets and five distance/similarity measures that have been most commonly used in text clustering.

Ray Hidayat: Generating Fast Automated Reports for the Farnsworth-Munsell 100-Hue Colour Vision Test

The greatest quality of the Farnsworth-Munsell (FM) 100-hue test is that it is able to indicate the presence of visual defects at early stages, so making it more widely available means more people can be saved from going blind or having their eyesight degenerate due to disease. However, the FM test is not used as often as its merits would warrant because of its main drawback - the extensive amount of time it takes to generate its report. We have developed a system that reduces the amount of time to generate reports for this test from 60 minutes to 4 minutes. The substantial increase in speed means this highly useful diagnostic test can be used more often, allowing for better detection of eye disease at early stages. Additionally the system generates statistical analysis of the results in accordance with Verriest norms, and has been used at Christchurch Hospital for several years.

Simon Ware: Modular Controllability Verification Using Language Projection

Model Checking is the task of searching the state spaces of finite-state automata to see whether they satisfy certain properties of interest. In many practical applications, the state space is much larger than can possibly be fit in the memory of a computer. One of the methods developed to overcome this problem and make it possible to verify large models is the so-called modular method. This paper presents a new algorithm which is capable of reducing the state space by orders of magnitude using projection. The new algorithm can be used separately or in combination with the modular method. It has been tested on a large set of real-world industrial models of very large size, and is capable of solving at least one model that has not been solved before.

Stefan Marks, John Windsor and Burkhard C. Wuensche: Evaluation of Game Engines for Simulated Clinical Training

The increasing complexity and costs of clinical training and the constant development of new procedures has made virtual reality based training an essential tool in medical education. Unfortunately, commercial training tools are very expensive and have a small support base. Game engines offer unique advantages for the creation of highly interactive and collaborative environments. This paper examines the suitability of currently available game engines for developing applications for clinical education and training. We formally evaluate a list of available game engines for stability, availability, the possibility of custom content creation and the interaction of multiple users via a network. Based on these criteria, three of the highest ranked engines are used for further case studies. We found that in general it is possible to easily create scenarios with custom medical models that can be cooperatively viewed and interacted with, though limitations in physical simulation capabilities make some engines less suitable for fully interactive applications. We show that overall game engines represent a good foundation for low cost clinical training applications and we discuss technologies which can be used to further extend their physical simulation capabilities.

Kathryn Hempstalk: You Are What You Type?

Being able to recognise a computer user by their typing rhythm is an effective way of protecting a computer from unauthorised use. Current techniques are accurate in restricted situations. In order to apply typist recognition to a wider range of scenarios, a dataset consisting of key press/ release data is required. With this data, it is possible to investigate methods of recognising a typist on a continuous basis. This paper discusses existing typist recognition techniques and the process of collecting a new dataset of typing data. Strategies to improve on the existing techniques using the new data are proposed. The work covered here was undertaken in the first 18 months of PhD research.

Qihang Huang: View-Oriented Parallel Programming on Multi-Core Clusters

Driven by the ever-growing demand for computing power, computers are becoming more and more powerful. However, in recent years, due to the physical limitations, this increased computing power does not come in the form of increased CPU clock speed, but in the form of more cores (processors) in a single chip die. Computer industry has started to use this new multi-core technology to massively produce systems for both stand-alone desktop PCs and high-end servers. In the near future, multi-core cluster will become one of the most economic supercomputer architectures. In order to utilize the full power of multi-core systems, some kind of parallel computing is necessary. However, parallel programming is notoriously known as a challenge job. This paper analyzes different parallel programming models, compares their strengths and weaknesses on multi-core based systems, and introduces an on-going project on providing a better parallel programming environment based on a novel View-Oriented Parallel Programming (VOPP) model.

Nuchjira Laungrungthip, Alan McKinnon, Clare Churcher and Keith Unsworth: Sky Detection in Images for Solar Exposure Prediction

This project seeks to determine the solar exposure at a location at any time of the day on any day of the year, using a new technique which involves image processing. A series of photos is taken from a location of interest and are then processed to separate the areas of sky from the rest of the image. The sunlight that will fall on the location from where the images were taken can then be calculated. Critical to the success of this project is the image processing technique to separate the sky from the rest of the image. This paper is concerned with finding a technique which can separate areas of sky for a number of images taken under different conditions.

Ke Geng and Gill Dobbie: Using Data in the Transformation of XML Queries

XML query transformation is a key part of XML query optimization. However most research into transforming XML queries is schema-based rather than data-based because of the highly flexible nature of XML documents. In this paper, we introduce an engine that we designed and built to test the effectiveness of different data-based query transformation techniques. The information is extracted from the data or XML documents, and with this engine, queries can be automatically transformed if they fall into particular categories which we have defined. Because this engine is developed independent of any particular XML database and different functions are realized by different modules, the engine is extensible and easy to adapt for different database systems.

Short Papers

Sean Gordon and Napoleon Reyes: A Method for Computing the Balancing Positions of a Humanoid Robot

This paper describes a method for balancing a humanoid robot that uses fuzzy logic to combine a relatively small number of base positions into a final position which is used to stabilize the robot. Although it was not possible to fully test the algorithm the premise that it relies most heavily on could be tested, and was found to be a premise that could be safely assumed if the base positions were properly calibrated.

Di Zhou: A Pattern-Oriented Interoperability Framework within MDA

Software systems typically comprise multiple components which are developed using different technologies, thus the interoperability that enables the components to interact despite differences in programming languages and execution platforms is becoming a central issue. This paper describes how to use Object-Oriented design patterns to derive a cross-platform interoperability framework for bridging the different technology specific models within the Model Driven Architecture (MDA). In addition, this paper presents a MDA based approach to applying the interoperability framework in an automatic fashion to generate a bridge between two platform specific models in a multi-agent system.

Rashina Hoda, James Noble and Stuart Marshall: Agile Project Management

As agile software development gains awareness and popularity in the software industry, it also continues to capture the interest of the research community. There are several topics within the agile software development area that demand deeper understanding and research. One such topic is ‘agile project management’ which relates to the management of software projects that are developed using various agile frameworks such as eXtreme Programming (XP) and Scrum. This paper outlines a proposed research on agile project management. In particular we hope to explore the role of the project manager, the process and problems of transitioning into an agile framework, and the management of outsourced agile projects.

Andrea Schweer and Annika Hinze: Combining Context-Awareness and Semantics to Augment Memory

People of all ages and backgrounds are prone to forgetting information, even about their personal experiences. Existing systems to support people in remembering such information either continuously record a person’s experiences or provide means to store and retrieve clearly defined, isolated pieces of data. We propose a new approach: combining context-awareness with semantic information. We believe this approach to be superior to the existing systems in certain types of situations.

Gregory A. Caza: Computational Model of Cognitive Development: Filtering Ambient Speech to Facilitate Word Learning

Infants manage to learn word meanings in very noisy environments. Despite an onslaught of many potentially confusing examples, language acquisition actually becomes more rapid around 18 to 24 months of age. During the same stage of life, a number of cognitive milestones are also reached. A review of developmental psychology and linguistics produces a theory that these parallel progressions may be linked by more than mere coincidence. As part of my master’s degree, I have developed a computational model to investigate language learning. It simulates a child who identifies communicative acts and then follows cues from a caregiver to disambiguate a single-word learning situation. Early results from the model are statistically similar to those observed in previous psychological experiments with actual children.

Halizah Basiron: Corrective Feedback in Dialogue-based Computer Assisted Language Learning

Computer assisted language learning (CALL) systems are used by people to learn a language. CALL systems have provided a number of advantages for language learning such as their ability to provide consistent and flexible corrective feedback during the learning process. This feedback is referred as information about what is ungrammatical or unacceptable in a target language. This paper presents a literature study on the field of corrective feedback and CALL and describes the future plan for my PhD research.

Gabi Schmidberger: Data Exploration Tools for Multidimensional Data

This paper presents new semi-graphical data exploration tools for multidimensional data. All new methods discussed herein are based upon histogram-like density estimation. The new graphics are coarse visualizations of multidimensional histograms. The histograms where built using a new histogram method. This method splits the range into bins of varying lengths in a way that the density in each bin is closest to univariate. In multidimensional space the algorithm finds bins that represent multidimensional areas of similar densities. This histogram method is used for simple representation tools which list the attributes of these bins and help the user to gain information about clusters and patterns in the data. A further section drafts the design for a graphical user interface which displays pixel graphics and comprises interactive functions for data exploration.

Reynaldo Giganto: Generating Class Models through Controlled Requirements

A software requirements document describes the functions of the system under development. Studies suggest natural language processing techniques can be used to automatically translate this document to object models such as UML class diagrams. However, some significant problems that arise from inherent ambiguity in language have yet to be overcome. This paper discusses these problems, shows initial experiment results and outlines further research plan to help solve them.

Matthew Jervis: Integrating Physical and Digital Workspaces

The paperless office, although much talked about, has yet to become a reality. Until then the office is divided between the physical and digital realms. The goal of my PhD research is to further investigate the divide between these realms and develop a system to bridge this divide. This paper presents some background to this field, outlines the work that has been done to date, and the future of my PhD research.

Lech Szymanski, Brendan McCane and Nathan Rountree: Maximum Margin Perceptron - Towards Optimal and Deterministic Neural Network Architectures

We outline our research for finding methods of engineering optimal neural networks with supervised training. The approach is to use the maximum margin theory to obtain a unique solution for given data in a similar manner to Support Vector Machines (SVMs). Our interest however, is in simpler alternatives to the mathematically complex SVM algorithms, more suitable for parallel processing. A number of existing methods for perceptron-type learning in the context of maximum margin separation is presented. Performance and suitability of those algorithms for our research is discussed.

Wilson Siringoringo: Minimum Cost Polygon Overlay with Rectangular Shape Stock Panels

Minimum Cost Polygon Overlay (MCPO) is a unique two-dimensional optimization problem that involves the task of covering a polygon shaped area with a series of rectangular shaped panels. This work examines the MCPO problem in order to construct a model that captures essential parameters of the problem to be solved by generic optimization algorithms. Three algorithms have been implemented to perform the actual optimization task: the greedy search, the Monte Carlo method, and the Genetic Algorithm. The results are presented to show the effectiveness of the implementation software under various settings. This is followed by critical analysis of various findings of the research.

Edmond Zhang: Mining Spatially Related Features for Object Recognition

This research will establish how local image feature frequency and geometrical relationship between image features (keypoints) can be used for more efficient and accurate image recognition. We will ask two important questions. 1)Is it possible to use machine learning to select a relevant set of keypoints for a particular recognition task? 2)Can a “higher level” descriptor consisting of multiple keypoints in some spatial arrangement be mined in order to enhance recognition performance?

Andrew McKenzie, Ray Hunt and Cong Huynh: NAT Traversal Techniques in Peer to Peer Networks

Peer to peer (P2P) networking has important applications in Internet telephony, online gaming, multimedia communication, instant messaging, file and workspace sharing which raises the need for robust and reliable NAT traversal techniques. This paper discusses current and developing techniques and challenges posed by NAT traversal in P2P networks. Initially Network Address Translation (NAT) detection is categorised and both UDP and TCP traversal techniques are discussed. Methodologies such as Relaying, Connection Reversal, and Hole Punching are then analysed. Finally the development of a testbed is described which can be used to evaluate NAT traversal techniques and to determine appropriate configurations in order to achieve P2P networking.

Stefan Mutter and Bernhard Pfahringer: Propositionalisation of Multiple Sequence Alignments using Probabilistic Models

Multiple sequence alignments play a central role in Bioinformatics. Most alignment representations are designed to facilitate knowledge extraction by human experts. Additionally statistical models like Profile Hidden Markov Models are used as representations. They offer the advantage to provide sound, probabilistic scores. The basic idea we present in this paper is to use the structure of a Profile Hidden Markov Model for propositionalisation. This way we get a simple, extendable representation of multiple sequence alignments which facilitates further analysis by Machine Learning algorithms.

Xianglin Deng and Chien-Wei Lee: Security of VoIP - VoIP Security-SIP Flooding and its Mitigation

Voice over Internet Protocol (VoIP) is an emerging telecommunications technology that is already shaping the future of telephony. VoIP uses the Internet Protocol (IP) to transfer voice data in packets as opposed to the traditional circuit-switched network used by the Public Switched Telephone Network (PSTN). Since VoIP communications are based on an open environment such as the Internet, they are can be exposed to the attackers. Moreover, the strict performance requirements of VoIP have significant implications for security, particularly denial of service (DoS) issues. Due to the popularity and the wide deployment of the Session Initiation Protocol (SIP), one of the standard signalling protocols for VoIP, the project focuses on the SIP INVITE flooding attacks.

Norhayati Mohd. Ali: Specifying Visual Design Critic Framework

This paper describes part of the work in the development of a generic visual design critic framework embedded within an end user oriented domain specific visual language meta- tool. The focus of the research is to expand further the capability of the Marama meta-tools by proposing a visual design critic framework that allows tool critic support to be rapidly developed in parallel with the tools. Critic tools detect potential problems, give advice and alternative solutions, and, possibly, automated or semi-automated design improvements to the end user. This paper is divided into five parts. The first part deals with the concept of a design critic. The next describes background and motivation for this research. The third part overviews the research approach and the following part describes the current proof of concept prototype of a visual design critic tool implemented using the Marama meta-tools. Finally the conclusion summarises the progress and future work plans of the research and also the potential contributions of the research.

Tom Botterill, Stephen Mills and Richard Green: Stereo Reconstruction for Visual Navigation

Visual Navigation is the key to enabling useful mobile robots to work autonomously in unmapped dynamic environments. It involves positioning a robot by tracking the world as it moves past a camera. This paper describes techniques we are developing to improve the accuracy of transformations calculated by matching sets of points between two views. This tells us how far the robot has moved. In the best contemporary Visual Navigation algorithms an approximate estimate of the transformation is improved using the Bundle Adjustment algorithm to incorporate all information from point correspondences tracked over multiple views. This iterative optimisation procedure is computationally intensive and prone to convergence to false minima however, so it is desirable to start as close to the true solution as possible. We compare existing algorithms used to extract a transformation for this initial solution and adapt them to provide improved performance when tracking 3d points from stereo cameras that have uncertain depth.

Ignas Kukenys and Brendan McCane: Support Vector Machines for Human Face Detection

This paper describes an attempt to build a component-based face detector using support vector machine classifiers. We present current results and outline plans for future work required to achieve sufficient speed and accuracy to use SVM classifiers in an online face recognition system. We take a straightforward approach in implementing our own SVM classifier with a Gaussian kernel that detects eyes in grayscale images, a first step towards a component-based face detector. Details on design of an iterative bootstrapping process are provided, and we show which training parameter values tend to give best results. Conclusions drawn from our work up to date are consistent with previous research and problems encountered are to be expected by anyone building an object detection system - SVM classifiers with large numbers of support vectors are slow and accuracy depends largely on the quality and variety of training data.

Quan Sun: The Life of a Session Result

An internet search result could be a web page, a document, an image or a video file returned by a search engine, which is a match for a user’s query. However, in some cases, we cannot locate the best set of results by a single query. That is, we find the best set of results that satisfies our information need through different query sessions. We therefore call the results returned by different queries - session results. This paper firstly discusses a graphical user interface that intelligently handles session results. Secondly, beyond the representational space, we use the concept of ‘world-line’, an idea borrowed from physics (theory of relativity), to study the change history of a session result.

Andrés Adolfo Navarro Newball, Geoff Wyvil and Brendan McCane: Towards the Creation of Londra: A Virtual Expressive Animated Dog

We aim to create a virtual dog head capable of displaying and animating facial expressions. The literature on virtual pets and facial animation in animals shows no work using anatomically based simulation and most existing systems attempt to give the animals human-like expressions. We will use a layered model for Londra’s head and connect it to a virtual dog body to enhance expressions with body posture. We have identified anatomical features that will provide a method of parameterization and we propose an anatomically based coding system to organise these parameters. We have developed an object oriented display program which we will use to visualise the model.

Sascha Rehbock and Ray Hunt: Trustworthy Clients: Architectural Approaches for Extending TNC to Web-Based Environments

Zhang Min, Clare Churcher and Alan McKinnon: Visualization of the Effect of Local Obstructions on Solar Exposure

This project seeks to create a sun path visualization. It will process a series of sun images to determine whether the sun is obstructed by nearby buildings, trees or other obstruction. The visualization will display the visibility of the sun from a particular location for any day or time and also provide solar exposure statistics.

Yulong Gu: What May Knowledge Management (KM) Technologies Do for Human Genetics?

This short paper describes key issues in human genetics knowledge processing. Identified problems are further discussed in the context of adopting Knowledge Management (KM) technologies. The conclusion of this work is that KM technologies may assist accelerating the genetics knowledge cycle - from genetic knowledge discovery to proper genetic healthcare.

Sam Bartels and Murray Pearson: Wireless Local Area Network Planning: An Overview

When planning a wireless local area network, there are design issues that need to be considered. In this paper, the fundamentals of planning a wireless local area network are introduced and discussed to highlight the requirements involved. Network constraints, as encountered in the physical environment, are discussed and their relevance to wireless network design is investigated. The paper concludes with an overview of wireless network planning solutions including commercial and free software, and an introduction to the author’s research.