| Computer Science and Software Engineering |  |
Other conference years: 2008, 2007, 2006, 2005
This conference is sponsored by
 |
Thursday 6th September (room 031) |
||
Session 1 – Keynote/Honours |
||
Chairperson:
|
||
8.50 |
9.00 |
Welcome
by the Conference Organiser |
9.00 |
9.30 |
Keynote:
Prof Peter Eades (
Humans Interacting with Information, Machines, Systems, and Humans |
9.30 |
9.50 |
Alexey Blinov: Digital music stand: an interaction study |
9:50 |
10.10 |
Toby Cox: Developing MAS using the MDA and Aspects |
10.10 |
10.30 |
Matthew Elliot: Tracking Fast Moving Objects in an Unconstrained Environment |
10.30 |
10.50 |
MORNING REFRESHMENTS |
Session 2 – Honours |
||
Chairperson: Ray Hunt |
||
10.50 |
11.10 |
Simon Fox: A framework for achieving collaboration within data consumer communities |
11.10 |
11.30 |
Keith Humm: Improving task switching interfaces |
11.30 |
11.50 |
Robin Muller-Cajar: Detecting advertisements in radio using Machine Learning |
11.50 |
12.10 |
Phillip Quinn: Understanding interfaces for command interaction |
12.10 |
12.30 |
Richard Viney: Rendering Real-time Soft Shadows with OpenGL 3.0 |
12.30 |
1.10 |
LUNCH |
Session 3 - MSc |
||
Chairperson:
|
||
1.10 |
1.30 |
|
1.30 |
1.50 |
Tobias Bethlehem: Theory of 3-4 heap |
1.50 |
2.10 |
Delio Brignoli: Performance estimation of 'distributed denial of service' mitigation techniques, a network simulation approach |
2.10 |
2.30 |
Cheng-Tse
|
2.30 |
2.50 |
Xianglin Deng: VoIP Security – Analysis, Testing and Mitigation of SIP-based DoS Attacks Specific to VoIP Networks |
2.50 |
3.10 |
Christiaan Gough: The effect of user familiarity on UI satisfaction |
3.10 |
3.30 |
AFTERNOON REFRESHMENTS |
Session 4 – MSc |
||
Chairperson:
|
||
3.30 |
3.50 |
Andrew McKenzie: NAT traversal by Peer to Peer networks |
3.50 |
4.10 |
Kapila Pahawalatta: Plant Species Biometrics using feature hierarchies |
4.10 |
4.30 |
Di Zhou: A Model-Driven-Architecture Based Approach to Developing Real-Time Multi-Agent Systems |
4.30 |
4.50 |
Xuan Gu: Incorporating Update Semantics within Geographical Ontologies |
4.50 |
CLOSE OF DAY ONE |
|
Friday 7th September (room 031) |
|||||
Session 5 – PhD |
|||||
Chairperson: Mukund |
|||||
9.00 |
9.20 |
|
|||
9.20 |
9.40 |
Jason Alexander: Document Navigation: Efficiency's Arch-Nemesis? |
|||
9.40 |
10.00 |
Oakley Buchmann: Usability Issues in Wearable Outdoor Augmented Reality |
|||
10.00 |
10.20 |
Adrian Clark: A Neural Network based system using scene analysis and adaptive filtering for Image-Based Object Registration |
|||
10.20 |
10.40 |
Phillip Cheng: Data integration and co-evolution: Data integration using ontology and the synchronization of schema evolution and ontology evolution |
|||
10.40 |
11.00 |
MORNING REFRESHMENTS |
|||
Session 6 - PhD |
|||||
Chairperson:
|
|||||
11.00 |
11.20 |
Isaac Freeman: Counting the costs: What kind of effort makes a novice into an expert? |
|||
11.20 |
11.40 |
|
|||
11.40 |
12.00 |
Hongzhi Gao: A Quantitative Comparison Research on Frame Level Background Subtraction Algorithms |
|||
12.00 |
12.20 |
Robert Grant: Constructing a 3D persistent local terrain map of an unconstrained environment for hazard detection |
|||
12.20 |
1.00 |
LUNCH |
|||
Session 7 - PhD |
|||||
Chairperson:
|
|||||
1.00 |
1.20 |
Mashitoh Hashim: Use of Efficient Data Structures for the Shortest Path Algorithm |
|||
1.20 |
1.40 |
Ray Hidayat: A trainable freeform surface-based recognition system for rigid three-dimensional objects |
|||
1.40 |
2.00 |
|
|||
2.00 |
2.20 |
|
|||
2.20 |
2.40 |
Moffat Mathews: A General Framework for Adapting Pedagogical Strategies in Intelligent Tutoring Systems |
|||
2.40 |
3.00 |
AFTERNOON REFRESHMENTS |
|||
Session 8 - PhD |
|||||
Chairperson:
|
|||||
3.00 |
3.20 |
MinKyung Lee: A Comparison of Two Display Types in a Simulated Multimodal AR Interface |
|||
3.20 |
3.40 |
Amanda Nicholas: Integrating Curriculum into Intelligent Tutoring Systems using Knowledge Spaces |
|||
3.40 |
4.00 |
Amali Weerasinghe: Developing a General Model for Supporting Self-Explanation for Intelligent Tutoring Systems |
|||
4.00 |
4.20 |
|
|||
4.30 |
Award ceremony – Staff Club |
||||
Retire to the Staff Club for end-of-conference function and presentation of conference prizes
Morning tea, lunch and afternoon tea will be provided both days on the tables in the basement level atrium. Drink vouchers and nibbles will be provided at the end-of-conference gathering and prize-giving - in the North Lounge of the Staff Club between 4:30 - 6 p.m. on Friday evening, 7th September.
Abstracts
Keynote Prof Peter Eades (
In this talk we discuss some technologies that support humans in their interactions with information, machines, systems, and other humans. We begin with Information Visualization, that is, the art and science of mapping abstract data into visual forms. These visual forms enable humans to understand large data sets (such as social networks, and biological networks). We give some interesting new and old examples, which show the successes and failures of computer-based Information Visualization. We describe some of the work of the "VALACON" project at NICTA, which investigates the visualization and analysis of large and complex data sets.
Next we describe the
"[braccetto]" project, which investigates how distributed teams
of humans can collaborate over a shared visual workspace. The [braccetto]
project is developing "mixed presence groupware", that is, tools
to enable partially co-located and partially remote teams to collaborate.
Finally we briefly describe the "HxI Initiative"; this is a major
initiative of the Australian Government aimed at technologies that aims to
increase
Honours:
Alexey Blinov: Digital music stand: an interaction study
The most important advantage digital music stands provide musicians with is silent turning of pages. Current commercial and research tools seem to overlook the aspect, rather concentrating on such features as annotating music and managing libraries of music scores.
Page turning techniques are limited to static page previews and the musician normally controls the system with two pedals: one to go to the next page and one to return to previous. This report investigates into alternative, animated page-turning methods and the ways musician could interact with the digital music stand. We explore whether animation provides the musician with visual cues or merely serves as a distraction; we also investigate alternative interfaces to the system, such as an on-screen menu, and how useful different musicians find them.
Toby Cox: Developing MAS using the MDA and Aspects
The development of Multi Agent Systems (MAS) can be assisted by the use of both the Model Driven Architecture (MDA) and Aspects. The MDA is a process that transforms a Platform Independent Model (PIM) of a MAS, into the required Platform Specific Models and their corresponding implementations. Aspects allow for the separation of Agency concerns (related to communication, collaboration etc) from core Agent behaviour (Beliefs, Desires and Intentions). This paper shall look at identifying Agency concerns within the MDA platform independent model (PIM), and using the MDA transformation process to create a PIM and PSM that incorporate Agency concerns as Early Aspects. We shall define MDA transformations so that Agency concerns can be represented as Aspects and introduced into MAS in a seamless and reusable manner.
Matthew Elliot: Tracking Fast Moving Objects in an Unconstrained Environment
In an unconstrained environment it is difficult to track the 3D flight path of a fast moving object with accuracy. There are two distinct problems that initially arise; the ability to locate the object in the scene amongst candidates and then tracking that object accurately. This research project aims to alleviate these two problems by introducing a new stereo vision algorithm which allows computation of a ground plane for reference. In the detection phase, anything moving in the scene is considered as a candidate object for tracking. The reference plane will allow these candidate objects to be filtered so that only realistic objects are considered. Once the object is located, stereo vision will afford the ability to accurately track the object’s 3D trajectory using the depth information gained from the scene.
Simon Fox: A framework for achieving collaboration within data consumer communities
Collaborative replication updates are an attractive property of Selective Data Replication in which data consumers cooperate to update their data set replications. Data consumer communities are implemented as groups of intelligent software agents who make decisions about when updates should occur. The software agent paradigm is suitable for achieving collaboration between individual agents, however some structured collaboration model must be followed. The theory of Cooperative Problem Solving (CPS) describes a theoretical model for achieving collaboration between a group of software agents. We present a framework, developed using the CPS theory, for achieving collaborative replication updates within a community of data consumers.
Keith Humm: Improving task switching interfaces
We use desktop computers today for increasing numbers of very different tasks. More often than not each of these tasks requires a number of different applications to be used. Interfaces to switch between these different applications and windows are difficult and cumbersome to use, often being unpredictable or difficult to predict when applications and window use changes. This project focuses on the development of an alternative, predictable interface to minimise the effort required to switch. However, in order to do this, user's task switching habits need to be understood. The second aspect to this project involves an empirical evaluation of switching habits in attempt to gain this understanding.
Robin Muller-Cajar: Detecting advertisements in radio using Machine Learning
We present an application that can recognise if the audio stream from a radio station contains advertisements. By filtering out key attributes both from the temporal and frequency domain of the stream, it is possible to distinguish between music and non-music using a multi-layer neural network. We also present a method that might be able to distinguish between advertisements and the radio DJ. This method, utilising hidden Markov models, has yet to be fully tested. Our application can separate music from other audio to a high accuracy, even when used on radio channels it has not been trained on, thus providing a first step towards advertisement free radio.
Phillip Quinn: Understanding interfaces for command interaction
Command interfaces have taken many different forms as they continue to evolve. Extensive research across several disciplines has developed formal theories on aspects of user interface design.
Although designers are not consistenly aware of formal theory, these theories are important to understand user behaviour and are useful in predicting and modelling user interaction and performance. My research examines the foundational theories of command interface design and interaction, and explores the design considerations and novel theories for interface navigation.
Richard Viney: Rendering Real-time Soft Shadows with OpenGL 3.0
Shadows convey important visual and spatial cues, and simulating realistic shadow interactions in real-time virtual environments presents a significant computational challenge. The upcoming release of OpenGL 3.0 includes new hardware and shader technologies that can be used to improve the quality and performance of existing real-time soft shadowing algorithms.
MSc students:
A common thread throughout all of learning is practice and within our everyday lives, human beings are constantly placed in situations where a task is to be performed and some feedback given. This is a very common learning scenario; as common as learning to ride a bike or learning by reading. Across many domains, Intelligent Tutoring
Systems (ITSs) are used to facilitate practice, providing a customized learning environment and personal tutoring experience for students to learn at their own pace through effective student modeling and feedback. Most current ITSs are built around cognitive learning theory and focus primarily on providing negative feedback or corrective feedback, facilitating learning by correcting errors. Research into the behaviour and methods used by expert tutors suggest that experienced tutors use positive feedback quite extensively and successfully. This paper investigates positive feedback; learning by capturing and responding to correct behavior, supported by cognitive learning theory. The aim is to develop and implement a systematic approach to delivering positive feedback in ITSs, in particular SQL-Tutor, a constraint-based tutor which tutors users in the design of Structured Query Language (SQL) database queries.
Tobias Bethlehem: Theory of 3-4 heap
As an alternative to the Fibonacci heap and a variation of the
2-3 heap data structure by Tadao Takaoka, this research presents the 3-4 heap data structure. The aim is to prove that the 3-4 heap, like its counterpart 2-3 heap, also supports n insert, n delete-min, and m decrease-key operations, in O(m + nlogn) time. These data structures have a wide range of applications and can by utilised in graph algorithms which use priority queues, such as Dijkstra's Single Source Shortest Path.
Delio Brignoli: Performance estimation of 'distributed denial of
service' mitigation techniques, a network simulation approach
In the past 10 years a number of mitigation techniques for DDoS on IP networks have been proposed, but their relative performance has not been measured in a simulated network environment. Our goal is to design a distributed network simulation framework and define effective metrics to estimate and compare the performance of existing or future DDoS mitigation techniques. The framework will provide facilities to generate simulated network traffic that resemble both legitimate and malicious traffic. It will also handle aggregation and analysis of results form multiple replications of the simulation across different hosts thus providing a way of comparing the performance of different techniques in a standardized environment.
Xianglin Deng: VoIP Security – Analysis, Testing and Mitigation of SIP-based DoS Attacks Specific to VoIP Networks
VoIP systems are subject to Distributed Denial of Service (DDoS) attacks during the initiation (SIP) phase. Use of a SIP server further implies a single point of failure under these attacks. DDoS attacks have been identified as the most significant risk to VoIP switch system performance. The proposed project will examine how the DDoS attacks affect the performance of the VoIP system, and verify how the current stateful session-tracking firewalls deal with these problem. Literature reviews have demonstrated a variety of DoS VoIP flooding attacks and proposed various mitigation techniques. In the research we will apply these ideas to SIP-based flood DDoS attacks, examine and select the best solution for mitigating SIP-based DDoS attacks.
Christiaan Gough: The effect of user familiarity on UI satisfaction
This research presents an improved model for predicting the effect of a user's familiarity with a user interface on their satisfaction using it. This may be used to evaluate existing interfaces or to predict the performance of user interfaces prior to construction. The results consistently show greater accuracy and consistency than previous models, across several different applications and task domains.
Cheng-Tse
This research proposes a new Haar-Disparity approach to overcome problems associated with tracking upper body pose without constraints on clothing, illumination and background. A cascade of boosted Haar classifiers is used to identify human faces in video images, where a disparity map is then used to establish 3D locations of detected faces. Based on this information, anthropometric constraints are used to define a semi-spherical interaction space for upper body poses. This constrained region serves the purpose of pruning the search space as well as validating user poses. The results suggest that Haar-Disparity improves on prior skin manifold tracking research by relaxing constraints on clothing, background and illumination. Performance of alternative camera positions and orientations were also evaluated.
Andrew McKenzie: NAT traversal by Peer to Peer networks
Network Address Translation (NAT) is a common feature in modern networks. Its presence, however, presents a major problem in establishing Peer to Peer (P2P) connections - used by applications such as Skype. In order to counter this problem P2P applications employ various NAT traversal techniques. This presentation examines the various types of NAT, and the techniques used by P2P applications to traverse them.
Kapila Pahawalatta: Plant Species Biometrics using feature hierarchies
The goal of the study is to develop a plant species
biometric which will be developed using both global and local features
of leaf images. Living plant recognition is a promising but challenging
task in the field of pattern recognition and computer vision. In recent
years, various approaches have been proposed for characterizing leaf images.
Most of them were based on a global representation of leaf peripheral with
Fourier descriptors, polygonal approximations and centroid-contour distance curve. Global
representation of leaf shapes does not provide enough information to characterise
species uniquely since different species of plants have similar leaf shapes.
Others were based on leaf vein extraction using intensity histograms and
trained artificial neural network classifiers. Leaf venation extraction
is not always possible since it is not always visible in photographic images.
To overcome such impoverished global representation of overall leaf shape,
this study will mainly concentrate on the extraction of local information
of the leaf contour, which is shape of the leaf margin, shape of the leaf
base and shape of the leaf apex. Contour detection and contour representation
techniques will adopt existing algorithms to extract biometric information.
The consequent biometrics will be tested on 41 common
Di Zhou: A Model-Driven-Architecture Based Approach to Developing Real-Time Multi-Agent Systems
In this research, the applicability of Model-Driven-Architecture
(MDA) based approach to developing Real-Time Multi-Agent Systems (RTMASs) is studied. The resulting system of this work is a RTMAS that assists container terminal (CT) operation manager(s) to enhance CT productivity through better utilization of the available resources, e.g. quay cranes (QCs), straddle carriers (SCs). The CT management is a complex task, which requires rapid and distributed decision making. Although RTMASs paradigm has been considered as an appropriate solution to this kind of complex task, the problem is that developing such a system is time-consuming and most programmers are unfamiliar with the process. MDA based technology can solve or simplify the problem by partially automating the system development. More specifically, the proposed approach applies multiple Unified Modeling Language (UML) profiles to modeling the CT operation manager(s) and various types of available resources as software agents with Real-Time performance constraints. The resulting RTMAS model is then transformed into one or more implementation technology specific models, each of which is in turn automatically transformed into source code.
PhD students
Jason Alexander: Document Navigation: Efficiency's Arch-Nemesis?
Document navigation is the act of moving within and between documents. In this research we are interested in understanding and improving navigation within electronic documents. Research to date has been focussed around AppMonitor, our software application for recording user actions in Microsoft Word and Adobe Reader. This seminar presents a brief overview of the AppMonitor logging system with a description of the results from a longitudinal study and a controlled experiment. The significance and consequences of these results are discussed.
Oakley Buchmann: Usability Issues in Wearable Outdoor Augmented Reality
The aim of my PhD is to identify and resolve usability issues in Wearable Outdoor Augmented Reality (WOAR). I will present the results of the last three years and will give an overview of the final experiment that I am currently preparing. This experiment will compare the performance of our WOAR system to that of a state-of-the-art industrial GPS stakeout system.
Adrian Clark: A Neural Network based system using scene analysis and adaptive filtering for Image-Based Object Registration
Typical image registration applications use a "one tool for all jobs" approach. Often algorithms with a narrow field of operation will be applied to a wide range of operating environments, with varied success. The aim of this research is to analyse the strengths and weaknesses of various registration algorithms so as to train a Neural Network to choose the best tool for the job. The neural network will also choose appropriate image filtering algorithms based on the registration algorithm chosen, and environmental conditions.
Phillip Cheng: Data integration and co-evolution: Data integration using ontology and the synchronization of schema evolution and ontology evolution
Isaac Freeman: Counting the costs: What kind of effort makes a novice into an expert?
Whenever we encounter a new user interface, we are cast into the role of novices: we spend more time thinking about the interface than about the task we are trying to perform. With experience, we gradually internalise the way the interface behaves and we are freed to concentrate on our goals.
User interface design has traditionally focused on making software easier for novice users, but this does not always help novices become experts. Requiring users to make a little extra effort can encourage them to internalise the interface, and become experts faster.
I present early results from an experiment comparing the effects of different kinds of effort -- mental, physical and temporal -- on the development of low-level spatial expertise.
Environmental science problems, in particular, epidemiological studies, are
generally investigated with statistical/geographical approaches; the use
of computer algorithms is yet uncommon. In this talk, how computer algorithms
can help increase knowledge about environmental science problems will be
discussed followed by example case studies such as introducing the use
of a decision tree algorithm to investigate a New Zealand biosecurity issue,
the sea container contamination problem (Ministry of Agriculture and Forestry),
and the K-maximum subarray to investigate two problems; the association
of air pollution and health in Christchurch (Environment Canterbury and
New Zealand Health Information Service) and spatial analysis of the biosecurity
control in weeds (AgResearch). From this study, the use of the computer
algorithm successfully provides practical knowledge and suggestions to
help for the future
Hongzhi Gao: A Quantitative Comparison Research on Frame Level Background Subtraction Algorithms
Frame level background subtraction algorithms are widely used in real time computer vision systems because of their simplicity and efficiency. Besides the ordinary reference frame subtraction algorithm and double difference algorithm, a hybrid algorithm: reference frame subtraction double difference algorithm, has been proposed recently. In this research, an algorithm of synthesizing video clips with known ground truth is proposed; then, these video clips are used to compare the three frame level background subtraction algorithms quantitatively.
Robert Grant: Constructing a 3D persistent local terrain map of an unconstrained environment for hazard detection
Hazard detection research has generally been constrained to detecting static objects on level floors in indoor laboratory environments. This research proposes an approach to terrain map generation which, unlike prior research, has the potential to be deployed outdoors by handling unconstrained terrains and environmental conditions. A 3D terrain map is calculated from multiple sensors filtering environmental conditions modelled by discrete and non-discrete particles, noise, illumination and motion blur. Hazards, detected from the persistent terrain map, are grouped into static and dynamic obstacles and surface gradient/traction. Initial research is focusing on creating an illumination invariant filter for video. By selecting optimal colour models and white balancing algorithms we can achieve video that stays consistent with environmental changes. This is complemented by hue variance prediction which detects noise caused when illumination levels change. In this way low sensitivity error can be easily adjusted for and research can produce better solutions without the consideration of environmental lighting conditions. Current research has been involved in analysing optical flow to infer three dimensional structure of the scene using only a single camera by calculating motion and ground plane information.
Mashitoh Hashim: Use of Efficient Data Structures for the Shortest Path Algorithm
In this paper, we propose to design a new data structure, called as a quaternary heap. This heap extends the trinomial heap invented by Takaoka, which supports a decrease key in O(1) time, and insert and delete-min operations in O(log n) time, both in the worst case. The proposed heap is specially designed to handle a special type of input operation pattern such as the increasing order or similar operations.
The new concept to be introduced is increasing the trunk size by one, which uses the quaternary linking, compared to the trinomial heap that uses the ternary linking. This new invented heap is aimed to allow an optimal implementation of the Shortest Path algorithm and many other algorithms. This heap will also support the existing total time of standard used operations better than the optimal way whose complexity is O(m + n log n), where n is the size of heap.
Ray Hidayat: A trainable freeform surface-based recognition system for rigid three-dimensional objects
Vision is an absolute necessity when it comes to interacting with the real world at any kind of complex level. The long-term goal of this project is to construct the algorithm which can recognise any rigid three-dimensional object, given training. To constrain the problem, the object must be opaque and non-reflective. The current plan to achieve this goal is to recognise the 2D surface features that comprise an object, identified by their shape and texture. These features will be identified by using the Canny edge detector, quadrature disambiguation and a variant of Gabor texture segmentation to locate the discontinuities that bound each 2D surface feature. Features will be recognised as objects through a partial aspect-graph matching approach, taking advantage of the multiple view hypothesis which has been experimentally proven as the method used by the human visual system. The result of this system will be a recognised object and its position and orientation. It is unlikely that this algorithm will be able to run in real-time on the computers of today, the true contribution of this work will be the creation of an object recognition algorithm of incredibly high accuracy, further opening up the possibility in the future of creating machines that interact with the world on a complex level.
There have been many changes in the face of computing in the past decade, one of which is undoubtedly the introduction of mobile computing. The ability to be connected to the Internet at all times with the best available access interface introduces the concept of Always-Best-Connected (ABC) services. However, with a wide range of access technologies, mobile-networked devices together with increasing demands of mobile users, introduces the notion of heterogeneous mobile environments.
The overall objective of the research is to provide a managed ABC services over underlying heterogeneous wireless and mobile platforms while maintaining negotiated security and Quality of Service in an open and non-proprietary environment. We have proposed a new layered-approach and its architecture to extend IETF PBNM model for centralised management of networks and present a generic framework for policy management in a distributed environment. We have also proposed a new model and its architecture of a policy-managed mobile client, which interacts with the available managed networks and provides seamless mobility to the mobile user. The presented performance analysis validates our architectural approaches.
Magic Lenses are an interface concept particularly well-suited to providing Focus and Context in interactive visualisations. The addition of a Magic Lens partitions the user’s view into a primary view and a secondary view seen through the virtual lens area. Both views can be configured independently to provide simultaneous distinct views of objects or information. Although they were originally designed for 2D workspaces, Magic Lenses have been applied to 3D virtual environments, and more recently, Augmented Reality.
This work explores the implementation, application and evaluation of Magic Lenses in Augmented Reality, a technology that overlays virtual objects on the real world in real time. Evaluations of 2D and AR user interfaces employing Magic Lenses are presented, along with current work including novel extensions to the technique
Moffat Mathews: A General Framework for Adapting Pedagogical Strategies in Intelligent Tutoring Systems
Human one-to-one tutoring is held as the benchmark by Intelligent Tutoring System (ITS) developers for the effectiveness of learning. Expert human tutors select from a range of pedagogical strategies adapting them to each student. This ability to draw from a pool of strategies empowers human tutors providing them with options for varying teaching situations. Current ITSs usually have underdeveloped pedagogical modules that contain one set of pedagogical strategies. This could be because ITSs are large, complex systems that take a large amount of time to author. Moreover, educational research lacks empirically evaluated models and teaching strategies. A possible reason is that variables in human teaching are difficult if not impossible to control.
This research examines the possibility of constructing a framework that governs pedagogy in an ITS. Ideally, the framework should be general enough to connect to the varied types of ITS, while allowing experts to add and modify their own strategies. It should also contain a default set of strategies derived from currently accepted learning theories and educational models. Having such a framework would greatly enhance the teaching abilities of ITSs making them more similar to human teachers, whilst allowing for controlled evaluations of various strategies.
MinKyung Lee: A Comparison of Two Display Types in a Simulated Multimodal AR Interface
Augmented Reality is a technology that allows the overlay of virtual imagery on the real world. Much of the current research in AR is about tracking or registration of virtual objects from videos of the real environment. However, AR also offers new possibilities for interaction techniques with digital content and so some researchers are starting to explore novel interfaces for AR. Many current AR systems adopt immersive Virtual Reality (VR) interaction techniques without any modification, but in immersive VR systems users are separated from the real world and so there are also opportunities for new interface methods that are unique to AR. In our current research we explore how multimodal speech and gesture input could be used in an AR setting. Although multimodal input has previously been explored in 3D graphics applications and immersive VR settings there has been little previous work on multimodal AR interfaces. We are interested in both how users will want to input multimodal commands, as well as how different AR display conditions will affect those commands. In order to explore multimodal interfaces in an AR setting we have developed a simulation system that combines vision-based hand tracking with simulated speech input and screen-based or head-mounted display AR output. We used the OSGART software library to implement AR view with realistic rendering. For the simulation, we have two different types of display devices, a monitor and a Head Mounted Display (HMD), and performed simple object manipulation and scene assembly tasks. The application is controlled by a hidden wizard using an external command console.
Amanda Nicholas: Integrating Curriculum into Intelligent Tutoring Systems using Knowledge Spaces
Currently, Constraint-Based Intelligent Tutoring Systems (ITS) store information about domain concepts, but there is limited knowledge about the curriculum of the domain. This research proposes to investigate the addition of curriculum to ITS using knowledge spaces, a mathematical theory first proposed by Falmagne and Doignon. The introduction of curriculum would allow problems to be presented to the student in an order that makes sense within the curriculum. If students need to know concept a before they can master concept b, the
system can now ensure that problems are asked in this order. Curriculum can also be used to determine appropriate feedback for the student, analyse their current knowledge state more quickly, and provide explicit instruction of concepts.
Amali Weerasinghe: Developing a General Model for Supporting Self-Explanation for Intelligent Tutoring Systems
We present a project with the goal of developing a general model of self-explanation (SE) support, which could be used in both well- and ill-defined instructional tasks. We have previously studied how human tutors provide additional support to students learning with an existing intelligent tutoring system. Although the tutors were not given specific instructions to initiate/facilitate SE, there were instances when SE support was provided. Analysis of these interactions indicates that they have helped the students to improve their understanding of database design. On the basis of these findings, we developed a self-explanation model, which we present in this talk . We used this model in a Wizard-of-Oz study, to provide adaptive SE support. The results show that students did learn the relevant domain knowledge. Human tutors mostly agreed with the interventions generated from the SE model.
NOTES FOR SPEAKERS
1. Please keep your presentation to the allocated time of 15 minutes.
2. Five minutes is allowed for discussion after your talk.
3. Speakers with PowerPoint slides - It is your responsibility to load them into the computer before your session begins.
4. A computer and data projector will be available. Please load your presentations onto it before the start of your session.
Prizes will be awarded for the
· best PhD student talk
· best Masters student talk
· best Honours student talk
The organisers are grateful for
the support of the following sponsors:
Allied Telesis
is a world class leader in delivering IP network solutions to the global
market place. We create innovative, standards-based IP networks that seamlessly
connect you with voice, video and data services.
Originally part of the DSIR, Allied Telesis Labs-based in Christchurch,
New Zealand-joined the international Allied Telesis Group in 1999, and is
now the largest research and development centre for the group.
As one of the world's leading producers of computer networking equipment,the group employs more than 3000 people worldwide.
Allied Telesis design and produce a broad suite of products ranging from powerful 10 Gigabit Layer 3 switches to media converters. We provide solutions for an impressive range of network applications - from individuals connecting home offices to the World Wide Web, to Internet Service Providers supplying targeted services to thousands of sites, more than one million customers worldwide have chosen our technology for their networks.
The success of Allied Telesis Labs is built on the skills of our talented employees, who enable the company to compete on an equal footing with the world’s largest communications and networking equipment manufacturers. www.alliedtelesis.co.nz
The Christchurch City Council
is one of the South Island's largest employers - a progressive local authority,
responsible for ensuring the continued successful growth and development
of one of
More than 2300 staff work for the Council across 60 locations around the city and Banks Peninsula. These include professional and administrative positions in core infrastructural areas such as water, waste, roading and parks; as well as jobs within the Council's broader activities including its library network, art gallery and recreation facilities.
The Christchurch City Council is an organisation committed to achieving sustainable outcomes for the community, environment and people of Christchurch and Banks Peninsula. By working for the Christchurch City Council you will have an opportunity to work on a wide range of projects providing you with opportunities to further develop your breadth of skills whilst contributing toward the development of our beautiful city and surrounding areas.
HumanWare designs and manufactures innovative technology for
people who are blind, deaf-blind, visually impaired and learning disabled.
Formed in 1988, HumanWare exports approximately 98% of its products worldwide
to more than 30 countries. Sales offices have been established in the
HP New Zealand is a company
that serves everyone from consumers to small and mid-sized businesses to
enterprises to public sector customers with an extensive portfolio of market-leading
solutions specifically designed to meet the needs of each customer segment.
HP Labs provides a central research function for the company which is focused
on inventing new technologies to improve our customers' lives, change markets
and create business opportunities. HP has a significant presence in all markets
we serve: Consumer - leadership in handhelds, notebooks, printers and digital
cameras for rewarding experiences. Small and medium business - market-leading
products, solutions and services for simplified ownership.
Jade Software Corporation (JSC) is a specialist technology services company with more than 25 years experience in helping organisations quickly respond to business, customer and supplier demands.
Headquartered in Christchurch with offices in New Zealand, Australia, the United Kingdom, and the USA. JSC develops and markets JADE™ – a high-performance development and runtime platform for delivering targeted solutions to complex business problems.
JSC provides packaged software solutions for the Logistics, Tertiary, HR Management, Governance, Risk & Compliance and the Case Management and Intelligence sectors. JSC also provides custom development and a 24x7 systems management service.
By combining our expertise and agile development methodology with the JADE™ technology platform, we’ve enabled customers worldwide to deliver business-critical information that provides strategic impact with the reliability and robustness that comes from a mainframe computing heritage.
More than 2,000 organisations across Europe, North America and Asia Pacific use JADE™ solutions to manage their specific information requirements. Our customers span multiple sectors, and are often leading organisations within their fields. http://www.jadeworld.com
Microsoft New Zealand is
working with industry, communities, and governments to conduct our business
in a way that helps advance social and economic well-being. Our efforts include:
Digital Inclusion& Education: Helping people and communities gain access
to the tools, skills, and innovations they need to reach their full potential.
Economic
Orion Health,
New Zealand 's leading software exporter, develops world-leading healthcare
software that improves people's lives. The innovative company delivers an
end-to-end information technology (IT) integration platform that brings together
clinical content to improve clinician effectiveness and the quality of patient
care. Orion Health's solutions have been trusted around the world for twelve
years. Led by founder Ian McCrae from its
As the
world leader in Electrical Distribution and global number 2 in Automation & Control,
Schneider Electric develops a full range of products, software, systems and
services for four core markets: Energy &
Infrastructure, Industry, Buildings, and Residential. Our lineup, comprises
global brands (Merlin Gerin, Square D and Telemecanique), market-leading
local brands (PDL) Schneider Electric employs 320 people in
Tait Electronics Ltd was formed in 1969 and is a leading
provider of radio communications equipment to 160 countries. The company’s
global headquarters is in
Tait’s product offering has grown
beyond its roots in conventional two-way mobile radio to encompass a wide
range of products and services. From portable and mobile radios, through
to wide area trunked radio network systems that provide integrated radio
and telephone services for both voice and data communications. Tait is
a founding partner of the
Trimble is a leading provider of advanced positioning solutions that maximize productivity and enhance profitability. Though best known for GPS technology, Trimble integrates a wide range of positioning technologies including GPS, laser, optical and inertial technologies with application software, wireless communications, and services to provide complete commercial solutions. Its integrated solutions allow customers to collect, manage and analyze complex information faster and easier, making them more productive, efficient and profitable. Trimble products are used in over 100 countries around the world. More than 2,400 employees in over 18 countries, coupled with a highly capable network of dealers and distribution partners serve and support our customers. For over 25 years, Trimble has created unique positioning products that help customers grow their business. Our portfolio includes over 700 patents and serves as the basis for the broadest positioning offerings in the industry. Trimble augments its organic product development with strategic acquisitions to bring the latest positioning technologies to a wider market. www.trimble.com