Waikato University crest

Department of
Computer Science
Tari Rorohiko

Computing and Mathematical Sciences

Seminars Presented in 2003

Events Index

Path Planning and Navigation in Uncertain Environments

Dave Ferguson
School of Computer Science, Carnegie Mellon University

 

Monday 22 December 2003

Addressing qualitative risks with software development impact statements

Professor Don Gotterbarn
Software Engineering Ethics Research Institute, East Tennessee State University
Friday 12 December 2003
Software developers and software engineers have been evolving and refining techniques to mediate risks of developing software products that meet the needs of their clients. The risks focused on include: missed schedule, over budget, and failing to meet the system’s specified requirements. In spite of this attention to risks and reducing these risks, a high percentage of software still fails, leading to software development being characterized as a “software crisis”.

A narrow approach to risk quantitative analysis and understanding the scope of a software project has contributed to significant software failures. A process is presented which expands the concept of software risk and failure to include the qualitative issues such as social, professional, and ethical risks that lead to software failure. Using an expanded risk analysis will enlarge the project scope considered by software developers. This process also is incorporated into a software development life cycle. A tool to develop Software Development Impact Statements is also discussed. The talk will illustrate this process showing work done for the UK government’s plans for electronic elections 2005.

 

What is a content management system?

Sally Jo Cunningham
Department of Computer Science, The University of Waikato

 

Wednesday 3 December 2003

Usability - the technology/knowledge worker interface

Kirsten Thomson
Department of Computer Science, The University of Waikato
Tuesday 2 December 2003
Kirsten has been given the above topic (and accompanying presentation blurb, see below) to present at the “KM NZ 2003: Driving and strengthening the knowledge driven enterprise” conference hosted by Marcus Evans in Auckland on December 8th and 9th.

Knowledge that has been captured is often not utilised because of unfriendly technology. Therefore, you much consider individual users and the tasks that they perform within an organisation which contribute to KM. As information and data becomes ‘easier’ to locate and apply to innovation, development, and business improvement, organisational acceptance of KM spreads.

  1. Improving communication and interactions within an organisation due to more effective mechanisms.
  2. Usability and organisational quality.
In this session, Kirsten Thomson from the Usability Laboratory from the University of Waikato will illustrate how the management of content and documentation with ease of use in mind has helped to complement the effective dissemination of information.

 

Some aspects about the architecture of a Data Mining program for a large amount of data

Wilhelm Steinbuss
Department of Computer Science, The University of Waikato
Tuesday 11 November 2003
Is a three tier architecture appropriate where the middle tier itself consists of three layers?

The three tiers are

  1. a browser for presentation,
  2. the middle tier as a J2EE Application, and
  3. a DBMS.

The middle tier consists of

  • the Data Mining application
  • a mapping application transforming the requests to the language of the underlying DBMS (eg ORACLE, DB/2, etc), and
  • an access layer.

 

Improving the Performance of Support Vector Methods

Professor Tom Downs
School of ITEE, University of Queensland
Tuesday 28 October 2003
This seminar will explain the basic operation of support vector machines (SVMs) for both pattern classification and for regression. It will also describe some variations on the SVM theme that have been developed in recent years to provide either simpler learning algorithms or improved generalization performance. Some examples of the application of SVMs to practical datasets will be given with performance compared to other leading techniques.

Finally the seminar will describe some recent work carried out at the University of Queensland which allows the exact simplification of the (sometimes quite complex) solutions provided by SVMs. An additional method, that has grown out of this technique, will be described that allows identification of redundant training data (ie data that contribute nothing to the learning process beyond what is provided by other data).

 

Virtual Circuit Switching in Ad-hoc Networking

Yen-Rong (Dan) Sun
Canterbury University, Christchurch
Tuesday 14 October 2003
An ad hoc network is a collection of nodes connected by wireless communication channels, forming a temporary network without using a pre-existing infrastructure or centralised administration. Traditionally, an ad hoc network uses datagram switching for transmitting a message that is many packets long. An alternative technique, known as virtual circuit switching, has not been tried. Two main challenges for implementing virtual circuit switching in an ad hoc network are: (1) finding a medium access control (MAC) protocol that supports ‘virtual circuit’ and (2) dealing with the rapid changes of network topology. A major advantage of using virtual circuit switching is its capability to provide Quality of Service during a communication session.

Ad hoc Virtual Switching Routing (AVSR) protocol is a cross-layered traffic control protocol developed to demonstrate virtual circuit switching in an ad hoc network. It is a reactive routing protocol running over a self-administrative time division multiple access MAC layer. The evaluation of AVSR shows it is applicable to implement virtual circuit switching in an ad hoc network, however it also indicates the scalability issue that is caused by the design of MAC layer. The conclusion of this research gives significant hints for future researches of virtual circuit switching in an ad hoc network.

 

Wise Up - Waikato Uni Advertising and Branding

Stephen Knightly
Marketing Director, The University of Waikato

 

Tuesday 14 October 2003

A-mediAS: An Adaptive Integrating Event Notification Service

Annika Hinze
Department of Computer Science, The University of Waikato
Tuesday 7 October 2003
Event Notification Services (ENS) inform about the occurrences of events that are of special interest for the service’s users, e.g. the publication of new Internet content. Upcoming systems for event notification, such as mobile traveller information systems and remote monitoring of commercial buildings, cover multiple applications and integrate event data from different sources. A main challenge is the evaluation of user profiles under different and changing application requirements. Of particular interest are (1) profile evaluation using differing semantics and (2) high filter performance under changing system load. This talk introduces the design and implementation of A-mediAS - an adaptive event notification system. A case study of a remote monitoring application shows the effective adaptation of evaluation semantics in A-mediAS. Different event filtering strategies are introduced and their adaptation to varying event and profile distributions are shown. Finally, future research plans are discussed as well as possible connections to research in the Department. The goal is to transfer event notification concepts into new application areas, for example, into Semantic Web, tourism and e-learning applications.

 

Back to Reality: Building Collaborative Tools for the Real World

Mark Billinghurst
Director HIT Lab NZ, Christchurch (http://www.hitlabnz.org/)
Friday 3 October 2003
In the past computer interfaces separated users from the real world. However in recent years new technology trends have allowed computers to enhance user interaction with their surroundings. These include:
  1. Augmented Reality (AR)
  2. Tangible User Interfaces (TUI)
  3. Perceptual User Interfaces (PUI)
In this talk we describe each of these areas and then provide an overview of the research that the HIT Lab NZ is doing in these fields. In particular we describe how that when they are combined together AR, TUI and PUI techniques can be used to significantly enhance face to face and remote collaboration.

 

Metaprogramming in Databases

Gottfried Vossen
University of Muenster, Germany
Tuesday 16 September 2003
The management of procedural data has regained interest in recent years, due to an increased exploitation of database concepts in novel applications such as the Web. However, formal foundations are still in their infancy. In this talk, we present several query languages for databases containing queries. The first, which is only briefly reviewed, incorporates reflection into the classical relational algebra, thereby increasing its expressive power considerably. The second, called the meta algebra, emphasizes type safety, and augments the relational algebra with three new operators for computing with relations in which not only data values, but also algebra expressions can be stored. The third, Meta-SQL, shows how easily the concepts of the meta algebra can be put to work in a database system supporting XML. Indeed, representing stored queries in XML, and using the standard XML manipulation language XSLT as a sublanguage, we show that just a few features need to be added to SQL to turn it into a fully-fledged meta-query language. This approach can even be implemented on any current database management system.

 

Appliance Design: a new discipline for pervasive computing

Professor Peter Thomas
UCL
Tuesday 26 August 2003
This talk will look at the emerging area of appliance design as an extension to current work in ubicomp, consumer experience design and traditional HCI. The seminar will outline basic assumptions, intellectual roots, current status and possible futures. Background: The Appliance Design Network at http://www.appliancedesign.org.

 

Participatory Usability: supporting proactive users

Dr Dave Nichols
Department of Computer Science, The University of Waikato
Monday 23 June 2003
After software has been released the opportunities for users to influence development can often be limited. In this talk I will review the research on post-deployment usability and make explicit its connections to open source software development. The costs and benefits of user involvement in usability activities are discussed. The issues involved in the design of end-user reporting tools are described with reference to crash reporting tools, the Safari web browser and a digital library prototype implemented in Greenstone.

The talk will cover the material for a presentation at CHINZ’03 in July.

 

Linking MEDLINE to the GeneOntology

Dr Tony Smith and Professor John Cleary
ReelTwo Ltd. and Department of Computer Science, The University of Waikato
Tuesday 17 June 2003
MEDLINE is a collection of some 13 million abstracts in the life sciences maintained by the NCBI. The Gene Ontology (GO) is a controlled vocabulary of thousands of terms used to describe the function of genes. We describe how machine learning technology was used to enable all 13 million abstracts to be automatically annotated with GO terms. The talk will look both at the underlying technology used to do this as well as the resulting system for organizing and searching MEDLINE.

The talk will cover material in two presentations being given at the Intelligent Systems for Molecular Biology conference in Brisbane at the end of June.

 

Audio interfaces for mobile devices

Professor Stephen Brewster
Department of Computer Science, Glasgow University
Tuesday 27 May 2003
The talk will describe some of the work we have been doing on using sound to improve usability in mobile devices, plus gestural interaction and evaluation techniques for mobiles.

 

Motion Planning in Dynamic, Uncertain Environments: A Bayesian Network Approach

Dr Michael Mayo
School of Information Technology, Bond University
Tuesday 7 April 2003
Motion planning is a combinatorial problem emerging from robotics and artificial intelligence. Given a robot (e.g. an articulated mechanical arm or a Sojourner rover-type machine) in a workspace that may contain obstacles, the problem is how to devise a plan for the robot to achieve some task whilst avoiding collisions with the obstacles.

It is well known that when the obstacles are polyhedral and motionless, the problem is computationally complex for all but the simplest of robots. Consequently, most efficient solutions to the motion planning problem rely on randomised or heuristic algorithms.

I will consider the much more difficult problem of motion planning when (i) the obstacles are dynamic (i.e. have varying velocities over time), and (ii) there is model uncertainty (the trajectories of the obstacles are initially unknown).

I propose to extend a class of randomized algorithm called Probabilistic Roadmap (PRM) Planners, by overlaying a Bayesian network onto a subset of the PRM’s vertices. This will allow the robot to compute firstly the partially safest path (as opposed to the shortest path) across the roadmap, and secondly, to decide whilst en route along a path whether to pause or proceed along “risky” portions of the path.

In this talk, I will focus on firstly the background to the algorithm, and then describe the algorithm itself.

 

GATE - A Software Architecture for Language Engineering

Mr Valentin Tablan
Department of Computer Science, University of Sheffield
Tuesday 25 March 2003
GATE comprises an architecture, framework (or SDK) and development environment, and has been in development since 1995 in the Sheffield NLP group (http://nlp.shef.ac.uk/). The system has been used for many language processing projects; in particular for Information Extraction (http://gate.ac.uk/ie/index.html) in many languages.

The presentation will include a concise overall description of GATE and a demo of GATE at work.

 

The PetaPlex Project

Robert M. Akscyn
President, Knowledge Systems
Friday 21 March 2003
The Petaplex Project was a research project funded by the National Security Agency to develop a digital library architecture scalable to 20 petabytes, throughput at the level of millions of accesses per second, and subsecond response. This talk will present the results of the prototypes constructed, lessons learned, and avenues for future exploration. Rob Akscyn has been President of Knowledge Systems for the past 22 years and has worked primarily in the fields of hypertext and digital libraries. He founded SIGWEB - ACM’s special interest group on hypertext, hypermedia, and the web in 1990. His company is the developer of KMS (Knowledge Management System). Formerly Rob was Senior Project Scientist in the Department of Computer Science at Carnegie Mellon University where he worked with Allen Newell and others on the development of early hypertext systems, including a 5 -year project to design and deploy what today is called an ‘intranet’ – aboard the USS Carl Vinson, a nuclear-powered aircraft carrier. Rob is visiting for a week (19-25Mar) in advance of a year’s teaching appointment at the University of Waikato beginning this July. Anyone wishing to talk with Rob while he is here should email him at rakscyn@hotmail.com.

 

Content-Based Multimedia Information Retrieval - Challenges and Opportunities

Dr Stefan Rueger
Department of Computing, Imperial College London
Tuesday 18 March 2003
This talk is about Content-Based Retrieval, ie retrieval of images, videos, music, spoken documents and text based not on manual annotation but on intrinsic data and automated extraction. Some of the challenges in multimedia data are given by (i) the semantic gap between low-level features (eg, texture and colour distributions) and high-level concepts (eg, a vase on a table); (ii) the variety of information need from the user: one person is intrigued by the graphic artwork on the vase, another by the age or type of the object a third by the composition of the vase and table in space etc; (iii) the variety of features within multimedia documents. This calls for interactive retrieval methods, the use of relevance feedback - explicit or implicit, flexible learning algorithms that not only adapt themselves to the user and their information need but also help bridging the semantic gap between low-level and high-level features. These methods, when integrated into digital libraries, will enhance their searching and browsing capabilities and give access through unconventional query methods such as sketching, humming and retrieving by examples of what is relevant. The second part of the talk is about current work at our multimedia information retrieval team at Imperial College London: Video Search and Summarisation, Music Retrieval, Relevance Feedback in Image Databases, and Information Navigation within a set of Documents.

 

Particle Filters In Robotics

Professor Sebastian Thrun
School Of Computer Science, Canegie Mellon University
Friday 7 March 2003
In recent years, there has been a fundamental revolution in robotics pertaining to the way we program robot software. Realizing the enormous amount of uncertainty faced by robots in the physical world, researchers have begun borrowing from the field of statistics. This presentation will introduce the audience to an emerging body of research on sequential Markov chain Monte Carlo techniques in robotics. In recent years, particle filters have solved several hard robotic problems. Early successes were limited to low-dimensional problems, such as the problem of robot localization in environments with known maps. More recently, we have begun to exploit structural properties of robotic domains, to scale particle filters to spaces with as many as 100,000 dimensions. The presentation will highlight applications of these techniques in subterranean, outdoor, and aerial robotics. Joint work with Michael Montemerlo (CMU), Daphne Koller and Ben Wegbreit (Stanford), and Juan Nieto and Eduardo Nebot (Univ. of Sydney). Sebastian Thrun is the Finmeccanica Associate Professor of Computer Science and Robotics at Carnegie Mellon University. His interests lie in the areas of robotics, computational machine learning, and human robot interaction.

 

Optimal Linear Estimation of Self-Motion: A Real-World Test of a Model of Fly Tangential Neurons

Dr Matthias O Franz
Department Of Empirical Inference for Machine Learning and Perception - Max Planck Institute for Biological Cybernetics
Tuesday 4 March 2003
The tangential neurons in the fly brain are sensitive to the typical optic flow patterns generated during self-motion. We examine whether a simplified linear model of these neurons can be used to estimate self-motion from the optic flow. We present a theory for the construction of an optimal linear estimator incorporating prior knowledge both about the distance distribution of the environment, and about the noise and self-motion statistics of the sensor. The optimal estimator is tested on a gantry carrying an omnidirectional vision sensor that can be moved along three translational and one rotational degree of freedom. The experiments indicate that the proposed approach yields accurate results for rotation estimates, independently of the current translation and scene layout. Translation estimates, however, turned out to be sensitive to simultaneous rotation and to the particular distance distribution of the scene. The gantry experiments confirm that the receptive field organization of the tangential neurons allows them, as an ensemble, to extract self-motion from the optic flow.

 

Is Robot Behaviour Chaotic?

Ulrich Nehmzow
Department Of Computer Science, Essex University
Tuesday 25 February 2003
Research in mobile robotics to date has, with very few exceptions, been based on trial-and-error experimentation and the presentation of existence proofs. Such existence proofs demonstrate that a particular behaviour can indeed be achieved, but not, how that particular behaviour can in general be achieved for any experimental scenario. We argue that the next step in establishing robotics as a mature discipline is the development of a theory of robot-environment interaction. Dynamical systems theory is one way to address this goal. The talk will present the argument for a theory of robot-environment interaction, and results of experiments conducted with a Pioneer II mobile robot at Essex University on the application of dynamical systems theory to mobile robotics. Ulrich Nehmzow is a Senior Lecturer in Intelligent Autonomous Systems and Cognitive Robotics in the Department of Computer Science at Essex University.

 

Online Communities: Participation, Lurking and Sucessful Online Communities

Professor Jenny Preece
Department of Information Systems, University Of Maryland, Baltimore County
Thursday 20 February 2003
Online community developers and managers are always worried about lurkers. They want them to “actively” participate by posting messages and chatting. Even the term “lurker” carries strong pejorative connotations. Yet many lurkers view themselves as community members who have altruistic motives for their behavior. In this talk I will discuss a study in which we examined lurking behavior in 100 online health support and technical support communities. We collected data using in-depth interviews and by logging the activity of the communities over a three-month period. I will also present some preliminary findings from a study in progress in which we used online questionnaires to study over 350 online communities from four different categories of communities. Together these studies present a rich picture of participation and lurking online. The findings pose some interesting challenges for those designing software for online communities and for those managing activity in online communities. The talk will end with a brief discussion of criteria for judging the success of online communities. Participants’ concepts of success often differ from those developers.

 

The Eyes Have It: User Interfaces for Information Visualization (Part 2: Hierarchical and Zooming Interfaces)

Professor Ben Shneiderman
Department Of Computer Science, University of Maryland at College Park
Tuesday 18 February 2003
Human perceptual skills are remarkable, but largely underutilized by current graphical user interfaces. The next generation of animated GUIs and visual data mining tools can provide users with remarkable capabilities if designers follow the Visual Information-Seeking Mantra: “Overview first, zoom and filter, then details-on-demand.” Then dynamic queries allow user control of widgets, such as sliders and buttons that update the result set within 100msec. Seven types of information visualizations (1-, 2-, 3-, multi-dimensional data, temporal, tree and network data) will be shown in examples for U.S. Census, personal photo libraries, time series searching, and genetic expression data. This talk (Part II) focuses on hierarchical data presentations with treemaps for stock market data (www.smartmoney.com/marketmap), production monitoring/product catalogs (www.hivegroup.com), and Census data. Zooming interfaces are especially suited for hierarchical data. Two personal photo interfaces will be shown: PhotoMesa (www.cs.umd.edu/hcil/photomesa) and PhotoFinder (www.cs.umd.edu/hcil/photolib). BEN SHNEIDERMAN (http://www.cs.umd.edu/~ben) is a Professor in the Department of Computer Science, Founding Director (1983-2000) of the Human-Computer Interaction Laboratory (http://www.cs.umd.edu/hcil/) , and Member of the Institutes for Advanced Computer Studies & for Systems Research, all at the University of Maryland at College Park. He was elected as a Fellow of the Association for Computing (ACM) in 1997 and a Fellow of the American Association for the Advancement of Science (AAAS) in 2001. Ben is the author of “Software Psychology: Human Factors in Computer and Information Systems” (1980) and “Designing the User Interface: Strategies for Effective Human-Computer Interaction” (3rd ed. 1998) http://www.awl.com/DTUI/. He pioneered the highlighted textual link in 1983, and it became part of Hyperties, a precursor to the web. With S. Card and J. Mackinlay, he co-authored “Readings in Information Visualization: Using Vision to Think” (1999). His new book “Leonardo’s Laptop” appeared in October 2002 (MIT Press) (http://mitpress.mit.edu/leonardoslaptop).

 

Online Communities: Designing Usability, Supporting Sociability

Professor Jenny Preece
Department of Information Systems, University of Maryland, Baltimore County
Tuesday 11 February 2003
Usability is about users’ experience at the human-computer interface. Sociability is about the social interaction between people via software. Human-computer interaction researchers have been primarily concerned with usability. Only recently has greater attention been given to designing for sociability due to the phenomenal increase in people’s social interaction via computers. This trend started in the 1990s with the development of Internet and the Web, which in turn gave rise to community interaction online. In this talk I will describe a method for designing usability and sociability for online communities. This approach relies on participative community-centered development in which key components of usability and sociability are identified and mapped to each other. Successful mappings depend on understanding group dynamics, trust and social capital within the online community. I will review our research and discuss work on group dynamics in detail.

 

The Eyes Have It: Useer Interfaces for Information Visualization (Part 1: Introduction and Multidimensional Data)

Professor Ben Shneiderman
Department of Computer Science, University of Maryland at College Park
Tuesday 11 February 2003
Human perceptual skills are remarkable, but largely underutilized by current graphical user interfaces. The next generation of animated GUIs and visual data mining tools can provide users with remarkable capabilities if designers follow the Visual Information-Seeking Mantra: “Overview first, zoom and filter, then details-on-demand.” Then dynamic queries allow user control of widgets, such as sliders and buttons that update the result set within 100msec. Seven types of information visualizations (1-, 2-, 3-, multi-dimensional data, temporal, tree and network data) will be shown in examples for U.S. Census, personal photo libraries, time series searching, and genetic expression data. This talk (Part I) focuses on multi-dimensional data in dynamic scattergrams (www.spotfire.com), geographic presentations in dynamic choropleth maps, and temporal data visualizations for patient histories, plus time series data applied to financial and genomic data. BEN SHNEIDERMAN (http://www.cs.umd.edu/~ben) is a Professor in the Department of Computer Science, Founding Director (1983-2000) of the Human-Computer Interaction Laboratory (http://www.cs.umd.edu/hcil/) , and Member of the Institutes for Advanced Computer Studies & for Systems Research, all at the University of Maryland at College Park. He was elected as a Fellow of the Association for Computing (ACM) in 1997 and a Fellow of the American Association for the Advancement of Science (AAAS) in 2001. Ben is the author of “Software Psychology: Human Factors in Computer and Information Systems” (1980) and “Designing the User Interface: Strategies for Effective Human-Computer Interaction” (3rd ed. 1998) http://www.awl.com/DTUI/. He pioneered the highlighted textual link in 1983, and it became part of Hyperties, a precursor to the web. With S. Card and J. Mackinlay, he co-authored “Readings in Information Visualization: Using Vision to Think” (1999). His new book “Leonardo’s Laptop” appeared in October 2002 (MIT Press) (http://mitpress.mit.edu/leonardoslaptop).

 

Myths and Potential: The Web is Dead, Long Live Services...Computing for the Masses

Scott Jensen
Interaction Design Advisor
Monday 3 February 2003
Scott has been doing user interface design and strategic planning for over 15 years. He was the first member of the System Software Human Interface group at Apple in the late 80s, working on System 7 and the Apple Human Interface guidelines. He then joined the original Newton team for 6 years as a programmer/designer. After Apple, Scott was a freelance design consultant, doing work for Netscape, Mayo Clinic, American Express, and several web start-ups. For the last 3 years, he managed DesignLab, the product design centre for Symbian. His team’s mission was to design, prototype, user test, and specify future mobile wireless products. He lead and coordinated dozens of design, focus group, and research projects. Scott has shipped a shrink wrapped spreadsheet, been a part of 2 Mac OS releases, 5 Newton product cycles, 4 commercial web site revisions, designed 2 completely different mobile phones, run dozens of usability trials and focus groups, has 5 patents granted and 23 in application. He has his Masters in Computer Science from Stanford University.

 

TINA and the Internet

Dr Peter Komisarczuk
Nortel Networks
Wednesday 29 January 2003
The Telecommunication Information Network Architecture (TINA) promised the convergence of Telecommunications and computing through the deployment of an all-pervasive Distributed Processing Environment (DPE) and architecture based on the Open Distributed Processing Reference Model (ODP-RM) and the Common Object Request Broker Architecture (CORBA) from the Object Management Group (OMG). On this TINA developed a network based intelligent, flexible and extensible services and management architecture, which in its later years tried to encompass the competing Internet infrastructure. This seminar looks at the work of TINA specifically related to the Internet and discusses some of the issues that caused TINA to not reach its full potential and discusses some of the current developments that make a TINA system more feasible today. In conclusion the seminar looks at how TINA lives on in various standardisation bodies today and how the concepts and the huge amount of research and development put into TINA is being re-used.

Peter holds an MSc in Modern Electronics from Nottingham University gained in 1984 and a PhD from the University of Surrey in the field of distributed computing, telecommunications and the Internet awarded in 1998. Over the last fifteen years he has been a lecturer at the University of Greenwich and has worked for Ericsson Ltd, Fujitsu Europe Telecommunications R & D Centre Ltd and Nortel Networks (UK) Ltd. in the areas of data networking and telecommunications on various international research projects, standardization activities and product related activities. These activities have encompassed research and development of network intelligence in the form of Telecommunication Information Networking Architecture (TINA) and various Internet and broadband access technology initiatives.


Events Index