Smart Internet Technology - Seminar and Talks Schedule 2002


The Smart Internet Technology Research Group has a weekly meeting where informal presentation, progress reports and discussions are held. There are also some open seminars scheduled.

These presentations are usually on Wednesdays at 11.00am in G92, Madsen Building in the Adolph Basser Seminar Roon - G92. The presentations scheduled to be given in 2002 are listed below:

Date Speaker Topic
Dec 25 to Jan 29 Recess
Dec 18 No talks scheduled - School of IT end of year party
Dec 11 Stephen Cross The SEI's Strategy for Improving Software Engineering Practice
Dec 3 Edo Plantinga Student Modeling in a Mathematices Tutoring System using a Genetic Algorithm
Nov 27 Daren Ler Developing a generic machine learning tool via the integration of ensembles and meta-learning
Nov 20 Bjorn Landfeldt Service Enablers for the Mobile Internet
Nov 13 Gerhard Fischer Knowledge Management - Problems, Promises, Realities and Challenges
Nov 6 Albert Zomaya Dynamic Location Management for Mobile Computing
Oct 30 Rafael Calvo The OACS web application framework
Oct 23 Sanjay Chawla A survey of spatio-temporal index structures
Oct 16 Masahiro Takatsuka An Open Component-Oriented Visual Programming Environment for Run-time Integration of Data Analysis and Visualization Tools
Oct 9 James Dalziel


Kelvin Lawrence
Interoperability and Integration for University Learning: The COLIS (Collaborative Online Learning and Information Systems) Project

Web Services, we've come a long way in a short time
Oct 2 Ken Williams A Framework for Text Categorization
Sep 25 Eric McCreath Advnacements with IEMS and the Daily Planner
Sep 18 No talks scheduled - SIT CRC Conference: doing business in a wireless world
Sep 11 Sarah Kummerfeld

Ken Lin
Meeting Personal Assistant

Applied Workflow Technology in the educational documents delivery to reduce information overload
Sep 4 Aaron Quigley Research Background and the FADE Paradigm
Aug 28 Kevin Carillo and Antoine Girard

Laurent Cimolino
Using RDF for Knowledge Representation in a Teaching Context

Verified Concept Mapping for Eliciting Conceptual Understanding
Aug 21 Sam Holden

Michael Chapman
Content Selection in SITS

Project Overview
Aug 14 General catch up session
Aug 7 Mark Assad

James Clark
Context Aware Framework Built Upon a Spread Toolkit

An Investigation into Intelligent E-mail Filtering Agents
Jul 31 Sarah Kummerfeld
Michael Till
Progress Report (in g74c)
Jul 24 Judy Kay, Bob Kummerfeld and Kalina Yacef Trip report
Jul 17 Kang Lee
Alex Lai
Thesis overview talk
Jul 10 Ken Lin,
James Clark
Thesis overview talk
Jul 3 Mark Assad,
Adam Hudson
Michael Chapman
Thesis overview talks
Jun 26 Terence Kam,
Andrew Lum
Thesis overview talk
Jun 19 General catch up session
Jun 12 Annie Ding,
Ehremin Avila
Thesis overview talk
May 29 Harry Mak Thesis overview talk
May 22 Kalina Yacef,
Bob Kummerfeld
Judy Kay
AH2002, ITS2002 practice talks
May 15 Antoine Girard,
Kevin Carillo
Jyot Boparai
A tool for teacher's feedback
Email as conversations
May 8 Derek Corbett,
Carolyn Leng
Wireless network simulator,
Workflow based teaching
May 1 John Bell Demo of learning object repository
Apr 24 Sacha Groves,
Sam Holden
Peer Review
SITS - scrutable teaching strategy
Apr 17 Daren Ler,
Laurent Cimolino
Generic machine learning framework
Concept mapping for building conceptual user models
Apr 10 Dr. Olivera Marjanovic Workflow technology in education
Mar 20 Smart Networks Presentation by U Wollongong partners - cancelled as speakers unavailable at short notice
Mar 13 Informal Welcome new postgrads
Mar 06 Practice talks Presentation of proposals for CRC funding round
Feb 27 Michael Dowman Cognitive Modelling of Colour Words: Language Acquisition and Language Typology
Feb 20 David Everitt Wireless Networks
Feb 13 Amanda Miller ++ group organisation and literature collection - discussion
Feb 6 Judy Kay Trip report - IUI, UK, Sweden


The SEI's Strategy for Improving Software Engineering Practice
Speaker: Stephen E. Cross
High-quality software, developed with predictable and ever improving performance, cost, and schedule is fundamental to success in today's global marketplace. The Software Engineering Institute (SEI) at Carnegie Mellon University in Pittsburgh, Pennsylvania (USA) was created in 1983 with the primary objective of helping others make measured improvements to their software engineering practices. The SEI's comprehensive programme of work is focused on the management and technical practices necessary to achieve this objective. The talk will provide a brief overview of the SEI and its programme of work.

Short Bio:
Stephen E. Cross is the Director and Chief Executive Officer (CEO) of the Software Engineering Institute (SEI). The SEI is a United States Department of Defense sponsored Federally Funded Research and Development Center (FFRDC) situated as a college level unit at Carnegie Mellon University. He was appointed to this position on November 1, 1996. He also holds an appointment in Carnegie Mellon's School of Computer Science as a Principal Research Scientist in the Robotics Institute. Dr. Cross is a member of the Air Force Scientific Advisory Board and is the past chairman of the Defense Advanced Research Projects Agency (DARPA) Information Science and Technology (ISAT) panel. He has published more than 50 papers on technology transition and the applications of advanced information processing technology. He received his Ph.D. from the University of Illinois, his M.S.E.E. from the Air Force Institute of Technology, and his B.S.E.E. from the University of Cincinnati. In addition, he is a graduate of the U.S. Air Force (USAF) Test Pilot School (Flight Test Engineer Program), the Air War College, and the National Defense University. In 2002, Dr. Cross was selected as a Fellow of the Institute of Electrical and Electronic Engineers (IEEE) and a Distinguished Alumnus of the University of Cincinnati College of Engineering.

The slides for the presentation can be found here.


Title: Student Modeling in a Mathematices Tutoring System using a Genetic Algorithm
Speaker: Edo Plantinga
The abstract for this presentation will be posted in the near future.

For more inforamtion on the content of this presentation, please follow the following link:
http://www.cs.usyd.edu.au/~edo/


Developing a generic machine learning tool via the integration of ensembles and meta-learning
Speaker: Daren Ler
Machine learning is typically employed via an algorithm that induces a single hypothesis, which may result in poor predictive performance. Furthermore, with the specialization of tasks, implementations of such algorithms typically result in the duplication of work and are subject to the expertise of the developer. This presentation will briefly look at two fields within machine learning that attempt to expand the limitations of a single fixed bias: ensembles and meta-learning. In particular, the talk will highlight the potential overlaps in the these fields and emphasize their use in the generation of a generic tool for machine learning.


Title: Service Enablers for the Mobile Internet
Speaker: Bjorn Landfeldt
The talk will be based on Bjorn's current research directions and in particular will look at providing the necessary support for applications to behave well in a heterogeneous networking environment with the focus on the emerging 3G networks.

Bjorn's web site can be found at:
http://www.landfeldt.com


Title: Knowledge Management - Problems, Promises, Realities and Challenges
Speaker: Gerhard Fischer
Traditional knowledge management (KM) approaches aim to archive information from the past so lessons will not be forgotten. This view implies that the information needs of the future will be the same as they were in the past. The focus of our approach is that knowledge is not a commodity to be consumed but is a collaboratively designed and constructed artifact.

Our perspective on KM is human-centered - focusing not on knowledge as information stored in repositories, but rather on a continual process in which knowledge is created as a by-product of work, integrated in an open and evolving repository, and then disseminated to others in the organization when it is relevant to their work. Our work is grounded in supporting creative design in different domains. Creative design, leading to large-scale design projects of long duration involving many stakeholders is one of the most challenging and most interesting domains for knowledge management.

The fundamental assumptions of our approach are:

In this talk I will review the problems and promises of knowledge management from this perspective, and illustrate how our work is creating conceptual frameworks and prototype systems in support of knowledge creation, integration, and dissemination.

Short Bio:
Gerhard Fischer (
http://www.cs.colorado.edu/~gerhard/) is a professor of Computer Science, a fellow of the Institute of Cognitive Science, and the director of the Center for LifeLong Learning & Design (L3D) at the University of Colorado at Boulder. Current research interests include new media supporting lifelong learning, human-human and human-computer collaboration, (software) design, domain-oriented design environments and universal design (assistive technologies). More information about the (L3D) center can be found at:http://www.cs.colorado.edu/~l3d/


Title: Dynamic Location Management for Mobile Computing
Speaker: Albert Zomaya
This talks presents a dynamic, individualized location update scheme that takes into account each user's mobility patterns. The mobility patterns are used to create individualized location areas for each user. The proposed scheme is flexible and can be used in network with arbitrary cell topologies. The scheme, along with other existing schemes is simulated using realistic users' mobility and call arrival patterns, and network topology. The simulated environment consists of 90 cells representing the geographical area of the San Francisco bay, and 66,550 mobile users representing the typical classes of users that are normally present in a real cellular network. Results show the proposed scheme gives lower overall signaling costs, resulting in savings on the limited radio bandwidth that may have otherwise been used for location updates and paging.


Title: The OACS web application framework
Speaker: Rafael Calvo
Application frameworks are software systems designed to increase reusability of design and implementation. In this presentation I will describe the OACS WAF, its architecture, its features and how it can be used in the University (for research and for T&L). OACS is an open source project. I will also describe the dotLRN Learning Management System built on top of OACS. dotLRN is being used (or developed) as the delivery platform at MIT Sloan School of Management, Heidelberg University, Berklee College of Music, Cambridge (UK), and UNED (Spain), Galileo University (Guatemala) and others.

Short Bio:
Rafael Calvo is a Lecturer and course developer in E-commerce at the University of Sydney - Sccool of Electrical and Information Engineering. He has a PhD in Artificial Intelligence applied to automatic document classification (e.g. web site classification). He has taught at several Universities, high schools and professional training institutions. He has worked at Carnegie Mellon University (USA) and Universidad Nacional de Rosario (Argentina), and as an Internet consultant for projects in Australia, Brasil,USA and Argentina. Rafael is author of a book and several publications in the field.


Title: A survey of spatio-temporal index structures
Speaker: Sanjay Chawla
The steady growth in mobile users has provided an impetus for research in database applications which can query and manage the location of objects over time. An important component of such applications are index structures that can efficiently retrieve moving objects. The research in this area has proceeded along two tracks. One can view a spatio-temporal index as a spatial index with a temporal extension. This view naturally leads to the extension of R-tree index, a familiar spatial index, to incorporate time. Another viewpoint is to model moving objects as a locus of algebraic curves. This transforms the problem into the realm of computational geometry. We will review both approaches and set the stage for future research in this area.


An Open Component-Oriented Visual Programming Environment for Run-time Integration of Data Analysis and Visualization Tools
Speaker: Masahiro Takatsuka
With massive increases in the amount of scientific data available, and corresponding increases in the complexity of analysis tasks, the kinds of analysis methods required by data analysis systems are changing remarkably.

Those analyses typically involve a wide range of activities such as statistical, computational and visual analysis. These activities are often carried out in the separate software tools or packages, and they are bounded by the intermediate results each software produces. The fundamental goal of this work is to provide an environment that integrates those software tools and packages in order to better integrate human-based and computationally-based expertise, and so ultimately improve scientific outcomes. The project addresses this challenging issue by providing the visual programming environment built on the basis of Component-Oriented Software Engineering.


Interoperability and Integration for University Learning: The COLIS (Collaborative Online Learning and Information Systems) Project
Speaker: James Dalziel
The COLIS (Collaborative Online Learning and Information Systems) Project is a collaboration of five Australian universities (Macquarie, Newcastle, UNE, USQ and Tasmania) together with a number of technology vendors (including WebMCQ, IPR Systems, WebCT, Fretwell Downing and Computer Associates). One of the central purposes of the project is to investigate and demonstrate strategies for the incremental development of levels of technical interoperability in learning space application integration.

The primary goal in 2002, as proposed in the Department of Education Science and Training (DEST) funding contract for the IMS Australia - Collaborative Online Learning and Information Services (COLIS) - Testbed, was to develop a distributed systems framework for online learning and information services which would demonstrate the following:

The interaction between the partners in the COLIS project has been a learning experience in its own right. Understanding the implications of interoperability in the learning/information space is still in its embryonic stages, particularly with reference to the management, delivery and use of learning objects. This workshop will explore the issues arising from the COLIS project, and the prospects for future development.

The slides for this presentation can be found here.

Title: Web Services, we've come a long way in a short time
Speaker: Kelvin Lawrence
In a few short years, a simple specification for the marking up data, namely XML, has spawned a number of exciting new technologies. Perhaps none more influential and with greater potential than Web Services. In his talk today Kelvin will look at where we have come from and where we are going with Web Services. Kelvin will discuss some of the current ongoing work in the industry to address the two biggest hurdles to widespread adoption of the technology. These are interoperability between implementations and the provision of adequate security such that secure web services transactions can be performed. He will also discuss a number of recent developments in the Web Services standards area and talk about his current work focus as part of the effort to make web services secure.

Short Bio:
Based in Austin, Texas, Kelvin Lawrence is IBM's CTO of Dynamic e-business Technologies. Kelvin is also an IBM Distinguished Engineer and a member of the IBM Academy of Technology. He is currently focussed on the advancement and deployment of Web Services technologies. One current focus is the addition of security technologies to Web Services. Kelvin is the co-Chair of a newly formed Web Services Security Technical Committee at OASIS. Prior to this Kelvin was IBM's CTO for XML Technology. In that role Kelvin was focussed on the development and deployment (in IBM products and services offerings) of XML technology (including parsers, style sheet engines and web services components).

Previously Kelvin was heavily involved in the prototyping and development of several projects based on Java (tm) Technology and more recently XML based technologies. Kelvin has also been heavily involved in the definition of internet standards and was the IBM representative to the World Wide Web Consortium (W3C) Scalable Vector Graphics (SVG) 1.0 working group, which defined the industry standard XML grammar for transmitting graphical documents over the internet. In the last three years Kelvin has worked extensively with XML and Java and other emerging technologies and has spent a lot of time teaching and lecturing on topics such as XML, Java programming computer graphics, user interfaces and e-business.

Kelvin has been very active in the XML community and he chaired the XML One conferences in London in 2000 and 2001 Kelvin's background and prior work focus is in the area of computer operating systems with a particular focus on graphical user interfaces and computer graphics subsystems. Kelvin has written numerous articles on a variety of related topics and has presented programming lectures all over the world and is currently coauthoring a book on SVG to be published later this year by Manning Publications.

Kelvin has filed in excess of 75 patents in areas such as computer operating systems, mobile devices, and human computer interfaces that have been filed in the United States and elsewhere. Kelvin has also been recently recognized as an IBM Master Inventor.

Kelvin has been with IBM in a variety of differing roles since 1986. He was born, raised and educated in England. He joined IBM UK in 1986 working at the Hursley Laboratory. He transferred full time to the United States in 1990. Kelvin holds an Honours degree in Computer Science from Brighton University in England.


Title: A Framework for Text Categorization
Speaker: Ken Williams
The field of Text Categorization (TC) emerged in the early '90s as an active field of academic study motivated by strong application needs. TC is a broad-based discipline, drawing inspiration from the Machine Learning, Information Retrieval, and Linguistics communities. It is seen as a key component of Knowledge Management in large businesses, as well as a helpful tool for individuals and smaller organizations.

The goal of my research is to create a reusable toolset for TC in the form of a software framework, drawing on the most applicable TC methods from published research and demonstrating their applicability in real-world applications. The framework is being implemented as an object-oriented hierarchy in Perl, which allows for extremely rapid application development or integration with existing applications. The seminar will consist of an introduction to the field of TC, a discussion of the design of the framework, and example applications.

The slides for this presentation are available at:
http://www.ee.usyd.edu.au/~kenw/


Title: Meeting Personal Assistants
Speaker: Sarah Kummerfeld
The organisation of meetings is a common task in the workplace. Tools have been developed to assist with meeting scheduling. However, these systems are quite inflexible, requiring each user to keep an electronic diary that is compatible with the system.

This talk describes a study we conducted to help understand what is a natural interfaces for meeting organisation and the implications for the design and implementation of a smart meeting personal assistant.

Title: Applied Workflow Technology in the educational documents delivery to reduce information overload
Speaker: Ken Lin
Workflow is a valuable technology. It is also a discipline, practice, and concept. It can be used to re-engineer system processes and increase efficiency. It can also provide Just-in-time resources delivery within an organization. In the University, a new staff or research student will receive hundred of pages documents within the first a few days. Most of the documents, he or she actually don't need to know at the beginning. My research will be applying the workflow technology in the educational documents delivery to reduce this kind of information overload.


Title: Research Background and the FADE Paradigm
Speaker: Aaron Quigley
In this talk Dr. Quigley will give an overview of his teaching, academic, and industrial research activities to date. He will also outline the plan for his role as a Senior Research Fellow within the Smart Internet Technology Research Group (SITRG). He will then describe one project in depth on his FADE paradigm for large-scale relational information visualization, clustering, and abstraction.

Algorithms for the visualization of abstract relational information, which tends to reveal that natural clusters in data tend not to scale well due to the high computational cost of determining element placement. This talk presents a series of efficient algorithms, for the two and three-dimensional layout of large graphs with thousands of elements. These algorithms are based on a hierarchical clustering of the nodes, codified in a graph model. This model is inspired by space decomposition approaches from N-body particle physics and plasma flow modelling. Within this talk Dr. Quigley will outline the results of two case studies, which demonstrate the models, measures and methods of the FADE paradigm. The large data sets used in these case studies are from the Matrix Market at NIST and resource flow graphs from the Bauhaus Project produced by partial RIGI analysis.

Short Bio:
Dr. Aaron Quigley will soon join the School of Information Technologies at the University of Sydney as a Senior Research Fellow in the SITRG. Dr. Quigley's research and teaching interests span issues in software engineering, data mining, web engineering, software visualisation, distributed systems, online societies and adaptive interfaces. He has written several peer-reviewed papers on his work, edited a book, and has a patent pending. Along giving many invited talks on his research work he was recently a program committee member for VisSoft02 and WGV02.

Dr. Quigley, a native of Dublin Ireland, completed his Ph.D. in 2001 in Computer Science while on faculty as an Associate Lecturer in the University of Newcastle, Australia. He received his B.A. (mod) and subsequent M.A. in Computer Science from Trinity College Dublin in 1995. He recently spent time working as a Postdoctoral Visiting Scientist at Mitsubishi Electric Research Laboratories (MERL) in Cambridge Massachusetts. Dr. Quigley has previously spent time working in Ireland, Japan, Germany, and the USA and is the recipient of an Institute of Engineers Australia Excellence in Engineering Award for the University of Newcastle.

His current web page resides at http://www.cs.newcastle.edu.au/~aquigley


Title: Using RDF for Knowledge Representation in a Teaching Context
Speakers: Kevin Carillo and Antoine Girard
Only a few applications have been designed in order to help the teachers to be aware of students' understanding and the problems they have met related to a specific course.

We will present our system designed to help the teachers discover problems their students are having as well as to gain insight into their developing understanding of their course. The work exploits ontologies to capture the content of student's informal messages between themselves in forums such as message boards.

After a brief overview of the system, we will present our work and explain advantages and possible enhancements.

Keywords: Semantic Web, Ontology, RDF(S).

Title: Verified Concept Mapping for Eliciting Conceptual Understanding
Speaker: Laurent Cimolin
This presentation will be based on the paper sent for the WBES conference in New Zealand.

Concept mapping is a valuable technique for education evaluation. This paper describes a system which supports teachers in creating concept mapping tasks which are intended to capture the student's understanding of the ontology of a small domain. A novel feature of our work is that the system verifies that the student intended the map elements that will be used to infer student understanding and misconceptions.

The presentation will also include a demo of the program.


Title: Content Selection in SITS
Speaker: Sam Holden
A look at the "instructional paths" generated via per-resource metadata in SITS.


Title: Context Aware Framework Built Upon a Spread Toolkit
Speaker: Mark Assad
Context aware systems allow the computer to be more involved with today's world. They will play a leading role in the development of ubiquitous, handheld, and wearable computer applications. In order for these systems to function, they must be able to commutation between each other in a common way. To this end it must be simple for application developers to construct compatible programs. The vast amount of context must also be communicated in an efficient way. This thesis attempts to describe a generic development framework for such context aware systems.

Title: An Investigation into Intelligent E-mail Filtering Agents
Speaker: James Clark
We are all aware of the growing tide of email messages arriving in our mailboxes. The email that you are reading now will be one of several that you may receive today, and one of the hundreds you might receive in a month. Increasingly, email is used as a communication medium for all aspects of our lives. The rate at which we are getting new mail will only increase. With such a large amount of email messages, many people prefer to sort their mail into categories, depending on the content. It is quite common for people to divide their correspondence between work items and personal messages. Most modern email client software allows the user to automate this process through the use of filtering rules. However, such rules can have holes. They cannot hope to predict the content of all the messages that will be arriving in our inboxes. It is from this need that current research seeks to create "Intelligent" email filtering agents; advanced machine learning algorithms to classify emails automatically. It is this area of research that I shall be summarising and presenting. From my review, I will draw some conclusions about what steps the current research could take in the future.


Title: Programmable Virtual Private Networking: Combining MPLS and Active Networks
Speaker: Adam Hudson
Virtual Private Networking is a valuable technique which allows for a network provider to partition their infrastructure into subnetworks, through the use of various tunneling protocols and security procedures. One such tunneling protocol in use today is Multiprotocol Label Switching (MPLS). While powerful and flexible, MPLS can be time consuming for subnetwork managers to configure, as they must ask the network provider to make any changes for them. Active Networking can be harnessed to provide an environment for subnetwork managers to directly interface with node switching hardware, thereby cutting these configuration times. This presentation will discuss how Active Networking can be combined with MPLS to create programmable VPNs and look at some of the advantages that this can provide over conventional VPNs.


Title: Digital Rights Management
Speaker: Terence Kam

We now live in the Internet age where information, services and even intellectual properties (software, books, music, video, photographic images and more) are delivered digitally. The use of the Internet is likely to grow, especially with the growth broadband, 3rd Generation (3G) mobile devices and powerful Personal Digital Assistant (PDA). With the growth of the Internet, a special problem arises. Digital information is merely bits and bytes that can be duplicated flawlessly and easily. As such, it is very difficult for authors to protect their intellectual properties from piracy. Without the ability to protect their digital property rights, there are no incentives for authors to produce intellectual properties. That in turn would be detrimental to society in the long run. Therefore, the solution to this problem is Digital Rights Management (DRM).

This project will investigate the issues confronting the consumers and owners of intellectual properties that use DRM. It will also look into which aspects of DRM that have the most relevance to computer security. Investigation will be carried out to find out to what degree are some of the above-mentioned aspects of DRM possible, comprehensive or desirable. Also, it will investigate and evaluate possible and existing techniques that can be used for some of the above-mentioned aspects of DRM. Finally, a very simple DRM system will be implemented to illustrate the concepts and ideas presented in this project.


Title: Mining of Microarray Gene Expression Data
Speaker: Ehremin Avila
The advent of microarray technologies have allowed biologists to detect and monitor simultaneously the activities of thousands of genes. This produces large amounts of data that need to be analyzed further and the challenge is how to interpret such massive and complex data. The aim of this project is to develop data mining methods and tools for analysis, interpretation and visualization of microarray gene expression data.

Title: Peer-to-peer wireless ad-hoc networks
Speaker: Annie Ding
Napster and Gnutella are very popular applications in the internet today. It is the high demand for information sharing which is pushing the development of Peer-to-Peer Network. Peer-to-Peer applications based on mobile ad hoc network can potentially provide much flexibility in network organization and dynamic delivery of services. Much of the research on peer-to-peer ad hoc networks concentrates on the architecture and routing optimization for these networks. The result will benefit the performance of the Wireless Peer-to-Peer network. This talk will survey the current state-of-the-art in research in peer-to-peer wireless ad-hoc networks.


Title: Movie Recommender Systems Using Learning for Text Categorization
Speaker: Harry Mak
This is an honours thesis project examining the potential of using reviews and summaries in text form to recommend movies to users. User preferences indicated by rating movies are the basis for a system to classify movies which the user has not seen. The implementation will draw its sources from text categorization, information retrieval and machine learning techniques and will use the Internet Movie Database (IMDB).

Keywords : Semantic Web, Metadata, Resource Description Framework, ontology.

Title: A tool for teacher's feedback
Speakers: Carillo Kevin and Antoine Girard
Web technologies provide a large range of tools and applications for teaching. Most of them allow teacher-student or student-student communication. However, only a few enable student-teacher exchanges that can provide the teacher with valuable feedback from his students. The Semantic Web allows the definition of precise knowledge representations. With the aid of metadata, it can support the use of relevant ontologies to create and update student profiles. The aim of this project is to retrieve and update student profiles from student inputs such as emails, newsgroups, chatrooms but also from quizes and exercises in order to build student models which the teacher could consult.

Keywords : Semantic Web, Metadata, Resource Description Framework, ontology.

Title: Email as conversations
Speaker: Jyot Boparai
Email has become a widespread means of communication and a most popular application of the Internet. The success of email has also encouraged its use as a medium for conversation and task management. This means it is being used for purposes beyond those it was intended for and this exacerbates information overload through email. Current techniques provide poor support for conversations conducted via email. This talk provides an overview of a method for managing conversations and explores its application to the a real world task of scheduling meetings.


Title: Wireless networks simulator
Speaker: Derek Corbett
Wireless networks encompass a large field of communications technologies and devices. This thesis involves simulating wireless networks specifically using third generation technologies such as WCDMA and cdma2000. By simulating wireless networks it is possible to trial the effects of changing different parameters without the cost of building the hardware and the network infrastructure to trial an idea. The use of a simulation environment overcomes many of the limitations found with analytical models. To this end a Java (OO) based simulation environment which can be used to examine, validate, and predict the performance of mobile wireless network systems will be built. By modelling the various components and their integration, the simulation environment will be able to accurately predict the performance bottlenecks of a multimedia wireless network running over a third generation network layer. To ensure the coherency of this environment with theory the data obtained from simulations will be validated against queuing theory which is the underlying mathematical model of the system.
Title: Learning objects repository
Speaker: John Bell

The Science Lectureships webforce project began in 2000 and is a three year collaboration between Monash, Queensland and Sydney Universities. Its goal is to develop teaching resources for the internet programmers. It funds the work of Tony Greening, Joe Thurbon and 51% of Kapila Wimalaratne. The speaker is John Bell, who is overall manager of the project and is located at Monash University. He has supervised the development of a repository which will support teachers in browsing and searching the resources developed in the programme.


Two talks this week giving an overview of student research projects

Title:Peer Review
Speaker: Sacha Groves

This talk will introduce the PEER REVIEW system. As a starting point, critical movie reviews will be analysed for information which can be used to develop a user model of the reviewer and their beliefs about the movie. The analysis will initially involve a manual study of the text and, with the aid of an interface that will be designed and implemented, collation of the user modelling information based on the views contained within it. This will then be coupled with conventional (bland) metadata about the movie. The resulting heightened (Judgemental) metadata created will then form the basis of a scrutable recommender which will potentially perform better than those which use only bland metadata. This theory will be experimentally tested and the system then extended to other areas such as teaching resources and peer reviews of journal papers.


Two talks this week giving an overview of student research projects

Title: Generic machine learning framework
Speaker: Daren Ler

Traditionally, when a machine learning algorithm is employed, it is designed and implemented to serve a specific task. Though these algorithms may be employed for a myriad of problems, they are often re-implemented to solve the problem at hand. The general consensus is that every problem is slightly different and therefore requires a slightly different implementation to achieve better performance. Though several tools have been developed to facilitate machine learning (such as C5, C4.5, ID3, Matlab's Neural Network Toolkit, etc), the amount of work that is typically duplicated to employ machine learning is significant.

This talk provides an overview to the conceptualization of a generic machine learning framework and attempts to identify the challenges involved in its construction.

Title: Concept mapping for building conceptual user models
Speaker: Laurent Cimolino

Since its development by Prof. Joseph Novak at Cornell University in the 1960s, Concept Mapping has been used for several purposes such as generation of ideas or design of complex structure. Another very interesting use is to represent the knowledge of someone in a specific domain (precise or not) with the aim to assess understanding and diagnose misunderstanding. This could be a new tool supporting the teacher-student relationship: the teacher builds the entire concept map of his domain and submits only a part (which can be nil) of it to the student who tries to complete the graph with his knowledge. The comparison of the two graphs will provide the understandings and also the misunderstandings of the student. In addition to concept mapping, I will present the program I am writing which allows the teacher to draw a graph and put rules on relations and student to assess their knowledge.


Title: Workflow technology in education

In the last couple of years, the area of flexible learning has become increasingly popular. However, the confusion remains about what all these terms such as flexible learning, flexible delivery, on-line delivery, e-learning etc. really mean and if and how they differ. As a starting point, this presentation adopts the following definition: "Flexible teaching and learning is that mixture of educational philosophy, pedagogical strategies, delivery modalities and administrative structures which allows maximum choice for differences in student learning needs, styles and circumstances" (Lundin, 1997). For many educators, flexible learning is seen as a direct consequence of the ways that information technologies are changing education. Unfortunately, that usually results in technology centered educational model where too much emphasis is placed on technology and delivery of instructions and content rather than on the learning process. At the same time, educational technologies are still lingering behind current developments in IT industry. They are still task-oriented rather than process-oriented.

Currently one of the most influential business information technologies is workflow technology. Workflows are process-oriented business information systems that offer the right tasks at the right point of time to the right people along with the resources needed to perform these tasks. In this presentation, the speaker will talk about her experience in implementation of an innovative workflow-supported learning environment and share the lessons learnt both from technical and educational perspectives.

About the speaker: Dr. Olivera Marjanovic is currently working as a Senior lecturer at the University of NSW within the School of Information Systems, Technology and Management. Olivera completed her PhD in Information Systems at the University of Queensland, Department of Computer Science and Electrical Engineering in the area of workflow technology. She also completed a GradCert Education at the same university in the area of flexible learning. Olivera has over 13 years of professional experience in the area of IT both in industry and at four universities. Her current research interests include: workflow technology, advanced learning technologies, e-services and knowledge management technologies.


CRC Practice Talks Bob Kummerfeld - Distribution and Management of Context Information (where this project has also been proposed by Chris Johnson and Rene Hexel

Joseph Davis, Judy Kay, Kalina Yacef - Just-in-time workplace training


Michael Dowman: Cognitive Modelling of Colour Words: Language Acquisition and Language Typology This research has investigated whether Bayesian inference can provide a plausible explanation of colour term acquisition and typology. Bayesian inference is a statistical procedure which allows inferences to be made on the basis of observed data, and the Bayesian model of colour term acquisition tries to determine the correct denotations of colour terms from examples of their use. This reflects the circumstances in which children must learn the colour terms of their own languages. The categories learned in this way display the prototype categories characteristic of colour terms, with colours closest to the centre of the category having the greatest degree of membership, while degree of membership decreases gradually towards the category boundaries. In order to investigate whether the model could also account for the typological patterns evident in the distribution of colour term systems throughout the world's languages, innate biases were added to the acquisitional model. It was hypothesised that the neurophysiologically determined innate red, yellow, green and blue foci would be more salient than other colours, and hence that language learners would be more likely to remember examples of these colours than non focal colours when learning colour words. Multiple copies of the model were then created to simulate a community of people, and these people would then take it in turns to name colours with which they were presented. Each time they would use the term which they thought most likely to be the correct name for the colour, given whatever colour words they had learned up to that point. This provided examples of the use of particular colour terms from which another person could learn, and this process was repeated over several generations of learners. Under these conditions, languages quickly emerge which consist of small sets of colour terms which roughly partition the colour space. Furthermore, the colour term systems reflect many of the typological patterns reported in the cross-linguistic literature on colour terms. For example terms with prototypes at the innate foci typically emerge very quickly, and terms without an innate focus tend to appear only in systems with larger numbers of colour terms. It is concluded that these simulations provide evidence that people learn colour term denotations using Bayesian inference, and that patterns observed in colour term typology may be a result of the effect of innate learning biases on the cultural evolution of languages.  
David Everitt: Wireless Networks This talk will give a broad survey of wireless networks past, present and future. The talk will be high-level, and will not assume any wireless knowledge. The should be a few pretty powerpoint slides.  
Judy Kay: Trip report - IUI, UK, Sweden   I will summarise some of the papers at Intelligent User Interfaces. (The web site has the papers and presentations.) This conference is highly relevant to the CRC. There were several papers about novel, ubiquitous computing interfaces as well as `natural' interfaces like sketching. Then I will briefly report on my short time in the UK visiting the group at Leeds and Susan Bull at Birmingham. Finally, I will give highlights of my amazing time in Sweden where I saw, at first hand, some wonderful approaches to PhD programmes and connected with Nahid Shahmehri's wonderful group. I will also briefly describe the aspects of Juha Takkinen's very interesting work on supporting users in email management.    

Back to top

Talks from 2001
Talks from 2000