2007 Seminar Series
Accent classification using support vector machines
Carol Pederson
Accent is the pattern of pronunciation and acoustic features in speech which can identify a person’s linguistic, social or cultural background. It is an important source of inter-speaker variability, and a particular problem for automated speech recognition. Current approaches to the identification of speaker accent may require specialised linguistic knowledge or analysis of the particular speech contrasts, and often extensive pre-processing on large amounts of data. An accent classification system using time-based segments consisting of Mel Frequency Cepstral Coefficients as features and employing Support Vector Machines is studied for a small corpus of two accents of English. On one- to four-second audio samples from three topics, accuracy in the binary classification task is up to 75% to 97.5%, with very high recall and precision. Its use with mis-matched content is at best 85%, with a tendency towards majority-class classification if the accent groups are significantly imbalanced. This compares favorably with a human listener study conducted using the same samples, which yielded an average accuracy of 92.5% (range 80) after an average of 7.7 seconds.
30 November
HTK: How does this speech recognition toolkit work?
Michael Norris
Michael has recently started as a research staff in the Thinking Head project within School of Information Sciences and Engineering.
This talk will be on uncovering the workings of a complex software toolkit “HTK” being used for development of speech recognition subsystem of the Think Head project. The HTK, Hidden Markov Model Toolkit is one of the most popular and complex software environment for the development of automated intelligent recognition systems.
9 November
Paradigms, Models and Features for Audio-Video Speech and Speaker Recognition
Michael Wagner
The Thinking Head Project is a multicampus collaborative project aiming at building an intelligent virtual embodied agent, or “avatar”. The University of Canberra human-computer communication group is constructing the input side of the avatar, including audio and video sensing followed by speaker-independent automatic speech recognition and the automatic recognition of speaker characteristics such as gender, age, dialect, accent and emotional state of the person speaking to the avatar. This seminar will provide a methodology for the dual audio-video speech and speaker recognition problems and will discuss potential solution paradigms, including suitable audio-video features and models.
Prof. Michael Wagner is currently the Head of the Discipline of Software Engineering in the School of Information Sciences and Engineering, and the Director of National Centre of Biometric Studies. He has worked in the fields of speech and speaker recognition, face recognition and behavioural biometrics since 1974. He was the principal research scientist of the TRUST Project from 1993 to 1996 and since 1996 has been the professor of computing at the University of Canberra. Michael’s current research includes face-voice recognition, liveness assurance, and multimodal fusion and secure protocols.
19 October
Auraya: Commercialization of Voice Authentication
Clive Summerfield
Dr. Clive Summerfield is an Adjunct Professor in the School of Information Sciences and Engineering and Deputy Director of National Centre of Biometric Studies. He is the founder and CEO of Syrinx Speech Systems, a pioneer in the development and delivery of speech recognition solutions with deployments in Australia and the USA. Dr. Summerfield is also the founder and CEO of 3SH with a specific interest in speech technologies for security and authentication. 3SH was the result of his 20 years experience in research, development and commercial deployment of voice authentication and speech recognition systems. Recently, Dr. Summerfield has also been awarded a major AusIndustry grant under COMET (Commercialising Emerging Technologies) Scheme. The media article about this grant is available online at the AusIndustry website.
12 October
Software Engineering : Knowledge and Performance
Narayanan Srinivasaraghavan
Software development is a knowledge intensive activity. The theme of this research is to explore the nature of the knowledge needed to conduct software development and the influence knowledge has on software development performance.
In the literature nine types of knowledge have been identified as being required for an individual to work effectively in a software project. These include configuration knowledge, social network knowledge, process and procedural knowledge, systems knowledge, business domain area knowledge, soft skills, technique skills and heuristics knowledge. The first of these, configuration knowledge, has to do with what a developer knows about the technical and architectural aspects of a systems development. This knowledge was the least explored in the literature so my original Professional Doctorate work concentrated on examining it and its contribution to developer performance.
During the work a number of additional theories of the knowledge required to perform various software development roles came to light. The upgraded Ph.D research now aims to explore more broadly and in greater theoretical depth the knowledge required to perform various software development roles . The research also aims to understand the contribution that increasing knowledge makes to perceived performance improvement.
A new theoretical framework has been developed covering theories on cognitive and social origins of knowledge and action.which account for the contributions of knowledge to individual performance. It is being tested using two research methods - first, a survey of developers to collect and analyse the knowledge requirements to perform their job, and, second, a case study of individuals in a software project setting. These methods test and refine the theory. The research will also produce an instrument for identifying, measuring, assessing and managing knowledge in practice that will help project managers to better understand the knowledge requirements of their projects and to determine recruitment and education of their developers..
5 October
Beyond the long-term mean: multivariate likelihood ratio-based FSR using F0 distribution parameters
Yuko Kinoshita
Long-term mean F0 (LTF0) is a popular parameter in Forensic Speaker Recognition (FSR). Its popularity probably stems from promising results in early SR research, together with its conforming to three of Nolan’s (1983) desiderata for FSR parameters, namely: robustness, measurability, and availability. On the other hand, many have noted that a single speaker can show large variation in this parameter from occasion to occasion, and even on the same occasion. Furthermore, Kinoshita (2005) demonstrated that because of its poor variance ratio (of between- to within-speaker variance), LTF0 shows very poor strength of evidence, typically generating Likelihood Ratios of effectively unity.
Forensic analysis of F0 has concentrated so far on its long-term mean and standard deviation. This paper explores the forensic discriminatory potential of combinations of parameters from LTF0 distributions. A multivariate likelihood ratio (Rose et al. 2006) is used as discriminant function on non-contemporaneous spontaneous speech samples of 201 Japanese male speakers from the Corpus of Spontaneous Japanese. Results are promising. With six features (mean, SD, kurtosis, skew, modal F0, probability density at mode) we obtain an EER of 10% (out of 80,400 different-speaker trials, ca. 72,000 are correctly discriminated, with ca. 182 out of 201 same-speaker trials correct). The EER is also extremely well calibrated.
Dr. Yuko Kinoshita is a lecturer and Japanese program convener in School of Languages and International Studies, Division of Communication and Education in University of Canberra. Her research interests are in the area of forensic speaker recognition. She is currently involved with forensic case work and research within National Centre for Biometric Studies.
28 September
Driving in the Dark: A Gentle Introduction to Nonlinear Control
Peter Vassiliou
Control systems are ubiquitous both in the build and the natural environment. Think of the control systems that operate in the plant and animal worlds, in ecosystems as well as in mechanical devises such as car engines and aircraft. Most of these control systems are of the nonlinear variety about which there is no general theory, in contrast to linear control systems which are generally well understood.
However, in recent years control theorists have managed to solve quite complicated nonlinear control problems, as for example the robotic backing and parallel parking of a car with 6 trailers in a confined space with obstacles! Such control systems are required for airport luggage carts, for example. In this talk I will describe, in simple terms, how this type of thing is accomplished. I will describe a simple nonlinear control problem involving a steerable, armless, wheeled robot and show how to make it do what I want! I will also describe a simple linear control problem and discuss its solution.
I will attempt to convey the ideas in broad terms, staying away from the technicalities as much as possible. In this sense the mathematical perquisites for following this talk will be very few.
Peter is a senior lecturer in mathematics and head of the mathematics and statistics discipline at UC. He has teaching interests in public key cryptography, coding theory and the mathematics that underlay these and related subjects. He has research interests in differential equations and in differential geometry and their applications. In the last few years he has developed a linearization theory for a general class of differential systems which includes nonlinear control systems as a special case. Peter enjoys walking in the sunshine along paths with gradients between -15 and +15 degrees, biographies, the history of science, general history and eating in fancy restaurants.
21 September
A user-centric approach to information security risk assessment
Dale Kleeman
This seminar will double up as the proposal seminar for Dale's PhD thesis, and the report on his OSP taken in semester 1.
All too often, information security practices within organisations are driven by IT departments and their associated information security functions, leading to what might be referred to as a security-centric approach to these requirements. He will argue that there is a case for the user and business areas of organisations to have a role in the formulation of informatiuon security policy and requirements, and when this occurs, it might be referred to as a user-centric approach to information security. Dale have developed an approach to facilitate user involvement in the formulation of information security requirements - this approach is based on ideas from control self assessment and soft systems methodology. The remainder of Dale's PhD will involve the testing of this approach through its application in a number of organisations.
14 September
The flood behind the Flood
David Clark
The story of the great flood in Genesis 6-9 is a foundational story in the scriptures of three religions.
It has variously been regarded as pure fiction, literal truth and many positions in between.The flood in the Genesis story, in both the English translation and particularly in the original Hebrew, has very specific and unique characteristics.There have been three major world wide flooding episodes in the 20,000 years since the height of the last ice age. One of these gave rise to a flood which is consistent with that in the Genesis narrative.
In this seminar David will briefly discuss some aspects of the flood story as told in Genesis. He will then summarise the astronomical and geological history of the earth since the last ice age as it relates to major flooding events. Out of this will come a candidate for the great flood of Genesis.
Dr. David Clark is a senior lecturer in the school of Information Science & Engineering. His interests include astronomy and bird watching, the former of which impinges on this seminar. He is married to Merilyn Clark who lived with the Genesis flood story for 7 years while doing her PhD, and who provided the Hebrew language background for this seminar.
7 September
Quantitative Decision Making
Alice Richardson
In this talk Alice will describe three collaborative research projects that all come under the broad title of quantitative decision making. The first uses data on NSW crimes and punishments; the second uses data on pathology test results; and the third uses data on animal populations, with application to teaching of introductory statistics.
Dr. Alice Richardson is a lecturer in the Mathematics and Statistics Discipline of the School of Information Sciences and Engineering. Her research interests include robust statistics, linear modelling and statistics education.
Dispute over a Grade: – Student Grievance Resolution Policy
Mohammad Yamin
Until now students could only appeal against a fail grade. From next year, students can appeal against any grade they get. University policy on grades in its current form is ambiguous on the method of calculating grades. Having been on the Students Appeals committee, Dr. Yamin has seen some interpretations of this policy which many of us might not like. This will be a short talk on this issue followed by discussion.
Dr. Mohammad Yamin is a lecturer in the Information Systems Disclipline of the School of Information Sciences and Engineering. His research interests include issues related to teaching and learning of information systems, in addition to the design, development and maintenance of large databases systems.
31 August
A collaborative multi-agent systems development methodology for software systems design
Ebrahim Fahad Al Hashel
Agent-Oriented Software Engineering (AOSE) is a modern approach that requires substantial research effort to make it available to the software industry. Agent-oriented paradigm provides humanoid type of concepts toward software systems problem solving. However, solving problem using the collaborative approach has been proved to be powerful techniques that software engineering aims to apply.
Collaboration is the key advantageous feature of Multi-agent system in comparison with all the other software approaches. Collaboration in multi-agent systems has the potential to decompose the complex task into several simple, low-level subtasks; it then dynamically forms an agent team and executes a plan to achieve a common goal. There are several theories, and models which have formed the bases of agent collaboration processes. In fact, each of these attempts has its own views and justifications as to what motivates the collaboration phenomena among agents and how the process is executed.
The existing agent-oriented software development methodologies, for example, Prometheus, Gaia, AUML, MaSE, PASSI, and Tropos are not able to engineer efficient multi-agent systems. Precisely, they are not able to design team formation processes at run time, or what is known as agent collaboration behaviour. This limitation is situated at the core of multi-agent software engineering practice and is reflected negatively on the system concept. As a result, the multi-agent systems can not function efficiently or exercises its powerful role.
The aim of this research is to improve and enhance agent software engineering by building a new multi-agent systems development methodology that has the capability to design collaborative multi-agent systems at run-time. The research will be conducted at two main stages:
1. Investigate the existing agent collaboration theories and principles and design an agent collaborations engineering model (ACEM);
2. Incorporate ACEM into Prometheus development methodology, to extend Prometheus to the new method MaSDM.
24th August
Abnormal Stock Market Behaviour and Communicative Practice in an Online Financial Community
John Campbell
The Internet plays an important role in keeping financial markets better informed by allowing private share investors access to information about exchange traded securities. Much of this information is provided from trustworthy sources including the official web sites of securities exchanges and home pages of listed companies. However, the Internet has also provided a medium that supports new forms of social networking such as the communities commonly found at financial Internet discussion sites. These communities are facilitated only by digital networks and would not exist otherwise. They can be an important source of collective insight on a broad range of topics relating to securities and securities trading. They empower private investors by providing access to information, and the knowledge and opinion of others. For many participants, these forums instil a strong sense of camaraderie and community. However, the potential exists for participants to be manipulated and exploited by dishonest individuals for personal gain. Such behaviour is not only damaging for the affected individuals but may also increase stock price volatility and create false market signals. This paper examines communicative practice in an Internet discussion site during a period of abnormal share trading behaviour. This is achieved by investigating an online discussion regarding the prospects of a listed Australian company exhibiting abnormal trading volume and stock price activity. The analysis shows that communication behaviours change significantly over time and are linked to changes in trading volume and share price volatility.
17 August
Mining the Archives: Towards a History of Geophysics in Australia
John Rayner
This talk examines three important phases in the development of geophysics in Australia. The first part relates to the Aerial Geological and Geophysical Survey of Northern Australia, (AGGSNA or simply "The North Australia Survey") which ran from about 1935 to the early 1940's. It was the first systematic study of this type undertaken in Australia north of the Topic of Capricorn and was a model for the later Bureau of Mineral Resources (BMR). The second part discusses the work of Australian geophysicists during World war II with a particular emphasis on the problems posed by magnetic mines and the counter-measures (degaussing techniques) that they and others devised to overcome the problem. The third part describes an extensive trip to the USA made by Harold Raggatt and my late father Jack Rayner from May to September 1945 to investigate the latest developments in geophysics, particularly with respect to oil exploration, and the structure of national geological surveys. The outcomes of this trip led directly to the formation of the BMR in 1946. The talk is based on Jack's papers and original source material from the Australian War Memorial and the National Archives.
Dr. John Rayner is an adjunct professor in physics, engineering and science communication in the School of Information Sciences and Engineering at the University of Canberra. He was a long time member of the staff of the University where he was the head of the School of ISE, and chair of the Academic Board. His research interests include plasma physics and its application to engineering problems, and science education. More recently as a visiting fellow in the Centre for the Public Awareness of Science at the ANU he has been working on a history of exploration geophysics in Australia.
10 August
Audio Programming Assistant
Philip Haines, Ngoc Khuu, Van Tieu-Vinh, and Ping Li
ISE team for the Imagine Cup 2007 is rehearsing their presentation before going to Seoul, Korea on Sunday 5 August for International Finals. They are keen to get feedback to allow them to refine their presentation so please come along and assist.Members of the Canberra Blind Society will come to evaluate the team project. Edward at UC will invite the media Australian and ABC News.
The aim of the project is to develop a system that gives blind and vision impaired users a simple and neat way to learn C# .NET programming and to share their learning with other users via the Internet. This project has opened a great possibility that allows blind and vision impaired users to become programmers in the future. Currently, blind and vision impaired people have little access to tools and assistance required for them to learn programming languages. The project aims to help blind and vision impaired people achieve equality of access and opportunity in information technology education that will ensure meaningful and equitable employment for their lives.
3 August
Face Tracking and Its Applications
Roland Goecke
Face tracking and the detailed tracking of facial features is a vital technology for many application areas, including for example human-computer interaction, affective computing, biometrics, and audio-video automatic speech processing. With recent advances in the area of face tracking, we are now on the doorstep of where many of these applications become more realistic in real-world scenarios. This talk will give an overview of current face tracking methods and talk about some of the applications.
Dr Roland Goecke is a Senior Research Scientist at Canberra's Seeing Machines company, which specialises in developing computer vision technology, in particular face tracking systems. He is also an adjunct researcher at the ANU. Before joining Seeing Machines, Roland worked as a researcher at the ANU and the NICTA Canberra Research Lab, as well as the Fraunhofer Institutes in Germany. Roland was awarded his PhD from the ANU in 2004 for work on audio-video automatic speech recognition, including a real-time stereo vision lip tracking algorithm.
27 July
UC Research Education Program
Joelle Vandermensbrugghe
The aim of this session is to discuss the research training sessions organised through the year for research sudents, where to find information on scholarships and degree requirements. Students will also be introduced to the on-line program linking students together, and which is being developed at the moment.
Joelle Vandermensbrugghe convenes the Research Education Program, a program designed for research students. The program coordinates training and centralises information to support research students. Joelle's background and current research activities are in the field of communication.
20 July
Quality, by design
Craig McDonald, ISE
Abstract: This seminar will review some ideas and models of 'quality' and its assurance. It will argue that 'quality' is an attribute, not a thing in itself - so QA should be seen as an explication of the design, execution and impact of a process or product. The seminar will draw on ideas of quality from the systems, software engineering and education disciplines. It will focus these ideas on the recent accreditation of ISE's courses by the Australian Computer Society (ISE's professional body) and on the future AUQA audit. The role of ISE's "unit design" in these quality assessments will be discussed. The seminar will be at least 60% open discussion.
13 July
Fundamentals of Signal Processing and Pattern Recognition for Face and Voice Authentication
Prof. Michael Wagner
Prof. Michael Wagner is the Head of the Discipline of Software Engineering and Director of National Centre for Biometric Studies in the School of Information Sciences and Engineering. Prof. Wagner’s HCC-NCBS research centre recently featured in an Australian Speech Science Technology newsletter available here http://www.assta.org/news/2007_1/Vol25No1.pdf
8 July
How the benefits of ITIL implementation can be measured
Partha Bhattacharya (Research Masters candidate)
Since early 2000 many Federal Government Departments, including University of Canberra are either implementing, or seriously considering introducing, the Information Technology Information Library (ITIL) Service Management Framework. DEWR, DFAT, IP Australia, ATO, DOFA, Austrade, DIMIA, Defence are the examples. Implementation of a Best Practice IT Service Management Framework such as ITIL is complex. If the complexity is not understood by the implementing agency the benefits claimed in the initial Business Case willnot be realized. IT Governance mechanisms within the Government agencies exaggerate successes and down play IT failures. The purpose of this study is to explore affective, organizational and environmental influencing factors that contribute to the development of operational performance improvement within, Department of Employment and Workplace Relations, a Federal Government agency. The study also seeks explanation as to why the improvements or benefits not realized as initially expected. Against this background this research will address the following key research question - how do we measure the benefit realization from ITIL implementation? The study will seek to understand the circumstances and settings in which ITIL implementation was undertaken within the organisation, how the use of ITIL processes contributed to overall performance improvement of IT Service Delivery and, based on those findings, come to a view on how the benefits of ITIL implementation can be measured.
29 June
Imagine Cup Winning Presentations
Smart Education
Jagdish Singh Mehra, Muhammad Meherban Arif, and Shafquat Zaman Khan
The purpose of this presentation is to provide a detailed picture of a project that provides a case study and demonstrates a possible solution to the Microsoft Imagine Cup theme. To address this theme, our project group known as Education Service Providers have developed a system called “Smart Education”. This system provides certain vital services to students from all backgrounds to make their educational pursuit a lot easier in terms of following lectures and tutorials with ease, managing their study time more efficiently, and in terms of understanding their study material when language is a barrier for them. There are other equally important services which are also highlighted. This presentation explains all the steps of the development process of our Smart Education System and explains what the system can and cannot do with careful consideration to different scenarios.
Audio Programming Assistant
Philip Allan Haines, Ngoc Thuy Duong Khuu, Van Tieu Vinh, and Ping Li
The aim of the project is to develop a tool that gives blind and visually impaired users a simple and neat way to learn C# .NET programming and to share their learning with other users. This project has opened a great possibility that allows blind and visually impaired users to become C# programmers in the future. Currently, blind and visually impaired people have little access to tools and assistance required for them to learn programming languages. The project aims to help blind or visually impaired people achieve equality of access and opportunity in information technology education that will ensure meaningful and equitable employment for their lives. This project Audio Programming Assistant (APA) aims to achieve the following objectives * Provide a voice-enabled programming tool that helps blind users design, edit, debug, and run C# console and windows application projects and provide voice output. * Provide a voice-enabled web browser to help blind users access any website on the Internet. Only links and texts on a web page will be converted to speech. Blind users can interact with any web page that requires users to enter some information. * Provide a website for blind users to download programming lessons to learn and to share their learning with other users by uploading text or voice messages.
22 June
Evolution of Contemporary RFID Systems
Son Le (Prof. Doc. candidate)
Radio Frequency Identification (RFID) technology provides a wireless means of communication between the tagged object and the reader. Successful simultaneous identification of a large number of tags by the reader in an efficient manner depends on a number of factors such as type of tags and readers, signal to noise ratio (SNR), distance, frequency and the physical environment (presence of water, metal, other interfering activities, etc.). This study we will develop a comprehensive and practical approach to help evolve RFID systems from just meeting mandate compliance to achieving business transformation resulting in more effective business operations and comparative advantages. It leverages our expertise and professional experience in three key related areas namely wireless, networking/communications and information technology including enterprise architecture.
15 June
3D Mapping of surface temperature using thermal stereo.
Surya Prakash (ISE PhD student)
A simple but novel technique to recover a 3D temperature map from a pair of calibrated thermal cameras is discussed. This method is also useful in estimating the depth of an object in a dark environment. Also discussed will be other alternatives for obtaining temperature maps on 3D surfaces including the method of space carving. The duration of the talk will be 30 mins followed by a ten minute demonstration with thermal cameras.
8 June
The Multi-agent Security Framework for e-health services
Sulaiman Rossilawati, ISE PhD student
Nowadays, the information and communication technologies are widely used in various applications for communications and sharing information. In the e-health domain, the Internet is used to deliver health services. Obviously, the information transmitted involves patients’ information that is considered as sensitive and the privacy of the information must be protected. The most popular communication choice is the Internet, as it offers cheap and worldwide coverage. There are many Internet-based technologies developed to facilitate the communication processes and health service deliveries such as web-application and wireless technology. However, as the technologies grow, the threats also grow in parallel. The threats put the patient’s information at risk if fell into the wrong hands. Therefore, it is important to make sure that the security technologies can cater for the privacy and security needs, whenever communications occur through the Internet. This research aims at developing a security framework that can cater for all security needs in e-health. Firstly, the users and the types of communications are described. Then, the multi-level communications approach for the users is introduced. Different levels of security mechanisms are used to cater for different levels of the users as well as flexible enough to facilitate different kinds of users’ needs in order to gain users' trust and confidence to communicate sensitive data through the network. Finally, the security requirements and the multi-agent based security approach are modelled and presented in the framework.
1 June
FUSION OF PROSODY, VOICE SPECTRUM AND FACE FEATURES FOR MULTIMODAL PERSON VERIFICATION
Mireia Farrus, ISE visitor
Multimodal biometric person recognition involves the combination of two or more human traits like voice, face, fingerprints, iris, hand geometry, etc., in order to achieve better results than using unimodal recognition. Multimodal person recognition systems normally use short-term spectral features as voice information; however, the performance of these systems is clearly improved when prosodic information is considered. Furthermore, the final performance results are highly dependent on the way the several biometric characteristics are fused and the type of the previously applied normalization.
25 May
Enterprise Architecture at ABS: findings from a "good" example
Neil Lynch, ISE
Enterprise Architecture (EA) is an area of research interest in ISE and of interest to many medium to large organisations, including UC. For example, the UC Accumen EA consultancy published in early 2006 developed an IT Enterprise Architecture for UC. Enterprise Architecture is seen differently by different stakeholders, not least of which is the difference in views between EA consumers, vendors and academia. There may be some common understanding of what EA might be but much that is written tends to be produced in isolation from other areas in organisations or, if produced by industry or academia, in isolation from other EA models and processes. Aside from the EA cognoscenti, even within organisations there may be little knowledge that an organisation possesses an EA or, if it does, what the EA means for that organisation, and even less understanding of the impact of that EA on work groups and individuals. Neil spent a large part of his OSP working at ABS (generally considered to be an 'exemplar' in it's use of EA) looking at the processes, models, and people, and from this research is proposing one means that may be useful in developing answers to some of these issues. The first part of this seminar summarises some issues surround this work and the processes used to gather data at ABS. The second part of this seminar is an abbreviated version of the research outcomes presented at the Australasian Conference on Information Systems in Adelaide in December 2006.
18 May
Testing Technology Adoption Theories: open source software adoption in the Australian public sector
Kavitha Gurusamy, ISE (PhD Confirmation Seminar)
The two main theories that have been proposed to account for the adoption of new technologies are Innovation Theory and the Technology Acceptance Model (TAM). The aim of this research is to test the validity of these theories in the domain of Open Source Software (OSS) adoption in the Australian Public Sector. This proposal begins with a review of theory and previous research on OSS, a review of current research in OSS and the creation of a theoretical framework identifying the full range of adoption factors reported in the literature. The core research question for this thesis is: Does technology adoption theory account for the adoption of Open Source Software in Australian Public Sector organisations? The research design is to conduct a set of exploratory case studies analysing OSS adoption in several Australian Public Sector (APS) organisations. The case studies will examine the specific factors at work in each case and compare them with those reported in the literature. An account will be given of why expected factors are not operational, or what new factors are operational in each case study. A pilot to confirm the research approach will precede the case studies. The outcomes of this research will reflect on and modify the current state of adoption theory. The practical contribution is to better inform policy making by providing knowledge on factors that influence OSS usage within APS organisations.
11 May 2007
On Extendable Software Architecture for Spam Email Filtering
Wanli Ma, ISE
The research community and the IT industry have invested significant effort in fighting spam emails. There are many spam email filters available and also in operation. Yet we are still inundated with spam emails everyday. This is not because the filters are not powerful enough, but because the filtering systems are not flexible enough to adapt the new development of spam techniques, such as HTML tagging, image based spam, and keyword obfuscating etc. In this paper, we propose to use dynamic multiple normalizers as the preprocessors for spam filters. The normalizers convert an email to its plain text format, called normalization. With the help of the normalizers, spam filters only need to deal with plain text format, which is what the filters are good at.
4 May 2007
TWM - a status report
Dave Davies, ISE
This will be a demonstration of the TWM Natural Language system and some discussion of recent developments and future plans.
Some recent issues in multivariate statistics and data mining
Shuangzhe Liu, ISE
This is actually my progress report on some selected issues I have recently worked on and been interested in. I will briefly answer such questions as:
1. What are the issues?
2. Why do we choose these issues?
3. How do we deal with these issues?
20 April
Framework for Integrated Tests
David Clark, ISE
The Framework for Integrated Tests (FIT) was written by Ward Cunningham. It is both a product and an approach. As a product it enables acceptance tests to be written independently of and applied automatically to test an application. As an approach, it fosters increased communication between testers, developers and customers.
FIT is specifically designed to test the problem domain (business rules). It does not test the user interface or usability or do stress testing. But what it does do, it does well. It can be used to test both calculations and processes. It is widely used in Agile methodologies, especially Extreme Programming.
This will be a gentle seminar that does not assume any prior knowledge or background in testing. I will demonstrate and discuss using FIT to test a variety of applications.
30 March
Open Source and Java in Canberra
Ross Ford, Federal Territory Manager, Red Hat (and JBoss).
Ross has 26 years in the IT industry - the last 19 in Canberra. Ross started as an Analyst Programmer, then a Systems Administrator. He has worked for vendors including NEC, Novell, IBM and Red Hat in Systems Engineering, Consulting and Sales roles. He has consulted and lectured across the Asia Pacific, US and South Africa. Ross has a B.Bus from Darling Downs Institute of Advanced Education - now USQ. He completed his final units at CCAE with Penny Collings in 1989
23 March
The missing link: introducing a developmental step into the UC academic integrity policy
In this seminar we report on a project that proposes an intermediary step in the UC Academic Integrity policy. We are proposing that a lecturer who has concerns that a student may be plagiarising (whether intentionally or unintentionally) should be able to direct the student to an interactive UC website called ‘Understanding plagiarism’. This site will have both generic and discipline-specific components. It will take students about one hour to read through the site and complete a randomised set of quiz questions. The aim of the site is developmental, and it will direct students to further sources of support and advice in the University and on-line. Lecturers will receive an email advising them that the student has successfully completed the quiz.
At present the site is still in its pilot stage, and we will welcome suggestions, advice and offers of collaboration from colleagues at the seminar.
9 March 2007
The Knowledge-based Economy and Knowledge Technology
2 March 2007
Spherical Harmonic Shading Models in Computer Graphics
Dr. Ian Lisle, ISE
Real-time 3-d computer graphics has always had difficulty drawing realistic shadows in a scene, especially the soft shadows characteristic of cloudy days or indoor scenes. In the last 5 years or so, techniques called Precomputed Radiance Transfer have been developed that go a long way towards solving these problems. A principal method for PRT is expansion of lighting and scene data in Fourier-like series of spherical harmonics. SH lighting separates calculations on the lighting environment from those on scene geometry, in such a way that almost all the hard work can be precomputed. The result is a runtime rendering of soft shadows that can dynamically adjust as the scene or light source is rotated. In this talk we demonstrate a fast, flexible implementation of spherical harmonic shading. Our implementation is able to dynamically change lighting environment and spherical harmonic degree, and we can even switch scenes by loading precomputed scene data from file. Contrasting different diffuse shading models shows the effectiveness of the technique, which in its most sophisticated version is even able to account for interreflected light in the scene.
This is joint work with MIT student Tracy Huang.
23 Feb 2007