Waikato University crest

Department of
Computer Science
Tari Rorohiko

Computing and Mathematical Sciences

Seminars Presented in 2004

Events Index

An evaluation of techniques for browsing photograph collections on small displays

Matt Jones and Steve Jones
Department of Computer Science, The University of Waikato
Tuesday, 14 December 2004
We evaluate techniques for browsing photographs on small displays. We present two new interaction techniques that replace conventional scrolling and zooming controls. Via a single user action, scrolling and zooming are inter-dependently controlled with AutoZoom and independently controlled with GestureZoom. Both techniques were evaluated in a large-scale, 72-subject usability experiment alongside a conventional thumbnail grid image browser. Performance with the new techniques was at least as good as that with the standard thumbnail grid, even though none of the subjects had prior experience with such systems. In a number of cases - such as finding small groups of photos or when seeking for images containing small details - the new techniques were significantly faster than the conventional approach. In addition, AutoZoom and GestureZoom supported significantly more accurate identification of subsets of photographs. Subjects also reported lower levels of physical and cognitive effort and frustration with the new techniques in comparison to the thumbnail grid browser.


Direct execution of Esterel programs on a reactive multiprocessor

Partha Roop
Department of Electrical and Computer Engineering, University of Auckland
Tuesday, 7 December 2004
Esterel is a language for modelling, verification and synthesis of embedded systems. Esterel supports the modelling of signals and sensors, delay, preemption, priority and combines these with a synchronous model of concurrency to effectively capture the features of embedded applications. Traditional approaches to Esterel compilation rely on mapping these features to software running on a conventional microprocessor that has no direct architectural support for typical features of a reactive language like Esterel. Moreover, simulating synchronous concurrency in a language like C (to which Esterel is compiled) through some form of scheduling code introduces many inefficiencies. Therefore, Esterel is not so effective for resource constrained embedded systems. This seminar will introduce an alternative approach to Esterel code generation through a reactive multiprocessor architecture called EMPEROR that has direct support for Esterel features at the instruction set. EMPEROR employs barrier synchronization to preserve Esterel concurrency. Benchmarking results (using Steven Edward's Estbench [IEEE TCAD 2002] suite) reveal significant performance and code compaction using the new approach. The proposed approach, thus, paves the way for resource constrained embedded application development using Esterel like synchronous languages.


An introduction to reinforcement learning

Kurt Driessens
Computer Science Department, University of Leuven, Belgium
Tuesday, 30 November 2004
Reinforcement learning is a sub-domain of Machine Learning that deals with unsupervised behavior learning. It is built on the same principles as animal training where the only information that is supplied to a reinforcement learning system is a quantitative assessment of its current behavior.

In this talk I will introduce the problem of Machine Learning, illustrate some of the most popular learning techniques for this kind of problem and discuss the most important issues of current research in the field.


Interfaces for staying in the flow

Associate Professor Ben Bederson
Institute for Advanced Computer Studies, University of Maryland, College Park, MD
Friday, 26 November 2004
The common notion of "staying in the flow" implies the ability to concentrate on a task, being immersed and engaged to the point where the perception of time slows. An artist, athlete or reader might not realize that half the day went by while working on something. And yet, most computer users would report that flow is something they rarely achieve.

Researchers have found that for people to stay in the flow, the task must be neither too difficult to discourage them, nor so easy that they become bored. A lack of interruption is also very important for staying in the flow. It has also been noted that flow is improved when users spend more time on the task domain, and less time on the interface domain (such as organizing, navigating, etc.)

In this talk, I will discuss how flow relates to user interfaces. By looking at interruptions, perceived time duration, and interface organization, I will show how interfaces can be designed to increase flow. To demonstrate these concepts, I will show a range of interfaces our lab has developed that we believe helps users to stay in the flow.


Battle of the bulge: the usability of fisheye visualizations

Carl Gutwin
Department of Computer Science, University of Saskatchewan, Canada
Tuesday, 26 October 2004
Although fisheye views and other distortion-oriented presentations have long been showpieces of information visualization, they have not found their way into common usage. A few minutes' experience with one of these systems suggests a reason why: although they are visually striking, they are also strikingly awkward. In this talk, I will review a research program that set out to determine whether fisheye views can be usable tools for viewing information and interacting with data, or whether they are a hopeless case, an infoviz trinket that should be returned to the mantelpiece whenever the real work starts.


Simulation of IPv6 Networks with OMNeT++

Ahmet Sekercioglu
Department of Electrical and Computer Systems Engineering, Monash University, Melbourne, Australia
Tuesday, 28 September 2004
In my presentation I will introduce our model development efforts on accurate simulation of IPv6 networks. Our simulation suite models several IPv6 related RFCs, and also MIPv6/HMIPv6 Internet-drafts. IPv6Suite also has an IEEE 802.11b model to simulate wireless LANs. I will then briefly talk about the collaborative research projects that use IPv6 Suite.


Reductions for machine learning

John Langford
Toyota Technological Institute at Chicago
Thursday, 9 September 2004
In complexity theory, reductions are typically used to prove “Problem X is hard”. In contrast, reductions in machine learning are a positive source of practical algorithms for solving a diverse set of learning tasks.

I’ll discuss a theoretical framework for analyzing these tasks which is “assumption free” in the same sense as the “online learning” model. Several of these theoretically analyzed (and optimized) algorithms have been implemented and tested yielding superior performance to other methods.


Digital music libraries

Dr David Bainbridge
Department of Computer Science, The University of Waikato
Thursday, 9 September 2004
In moving from physical to electronic form, digital libraries change the rules over how documents can be stored and retrieved — even what constitutes a document! This opens up new possibilities that for the past decade have been explored by digital library research initiatives around the world. This talk will summarise the work undertaken by the department of computer science in the area of digital music libraries.

The work encompasses handling different forms of music such as sheet music, symbolic representation, and audio performances; novel ways to access the information such as locating a musical score through a user singing a fragment of remembered melody as their query (in addition to supporting searching by textual bibliographic information). Alternatively, browsing through an ever-changing collage of images associated with music works can support a serendipitous form of access, akin to spotting an interesting book on a library shelf. Combining heterogenous forms of music also presents challenges that need to be overcome. During the course of the talk, a variety working examples will be demonstrated that encapsulate the ideas we have been developing.


Shape-based robot mapping

Diedrich Wolter
University of Bremen, Bremen, Germany
Tuesday, 17 August 2004
A novel geometric model for robot mapping suited for robots equipped with a laser range finder is presented. The geometric representation is based on shape. Cyclic ordered sets of polygonal lines are the underlying data structures. Specially adapted shape matching techniques originating from computer vision are applied to match range scan against the partially constructed map. Shape matching respects for a wider context than conventional scan matching approaches, allowing to disregard pose estimations. The described shape based approach is an improvement of the underlying geometric models of today's SLAM implementations. Moreover, using our object-centered approach allows for compact representations that are well-suited to bridge the gap from metric information needed in robot motion and path planning to more abstract, i.e. topological or qualitative spatial knowledge desired in complex navigational tasks or in communication.


Data mining case studies

Associate Professor Kate Smith
Monash University, Melbourne, Australia
Tuesday, 17 August 2004
This seminar presents a series of research and consulting projects that all adopt an holistic approach to data mining. Both directed and undirected knowledge discovery methods are employed in an effort to understand the patterns in the data and study their implications. Case studies focus on: identifying high-value and loyal customers in a retail chain; modelling the relationship between problem characteristics and the suitability of data mining algorithms; identifying rules for risk assessment of prostate cancer patients.


AI challenges in Web information retrieval

Mehran Sahami
Google Inc. and Stanford University
Tuesday, 10 August 2004
Search engines have become ingrained as one of the critical applications on the World Wide Web. Information retrieval, especially in the context of the Web, presents a host of challenges that must be addressed in order to better help people find relevant information in a growing sea of text. Such challenges include not only important issues in building large, scalable systems, but also providing intelligence to these systems to sift, organize, and present relevant information to users. In this talk, we provide a brief background on information retrieval, and then look at some of the challenges faced in applying IR to the Web. We will look specifically at how these challenges are addressed at Google, focusing on how AI technologies have been used in a variety of situations to improve users' access to information. We will also discuss Google's goal of continually improving web information retrieval through the deployment of a variety of technological advances in information analysis, understanding, and retrieval.


Indexing and retrieving visual information

Jesse Jin
School of Design, Communication and IT, University of Newcastle, Australia
Tuesday, 10 August 2004
One of the major difficulties in multimedia databases is indexing and retrieving visual information. The vectors generated from the visual features are usually in high dimension. Due to "the curse of dimensionality", it is not possible to efficiently search the exact k-nearest-neighbors of a point in a high dimensional space. There is also an issue of "semantic gap". This seminar reviews high-dimensional retrieval by examining the chronological evolution of various indexing techniques. It then presents solutions to important aspects of high dimensional index and retrieval, namely indexing structure, varying distance metrics, dimension reduction and visual vocabulary. Some creative thinkings are presented.


Robotics Group Research

Dr Margaret Jefferies, Dr Michael Cree and And Dr John Perrone
Departments of Computer Science, Physics, and Psychology, The University of Waikato, Hamilton
Tuesday, 3 August 2004
This seminar will give an overview of the current research being undertaken by the robotics group. We will cover our work on the Simultaneous Localisation and Mapping (SLAM) problem ie. estimating the robot's location while at the same time building a map of the robot's environment. We are currently investigating the use of landmarks to recognise when the robot is revisiting a place it has been to before. Margaret will discuss our approach using 2D landmarks computed from laser range data. Michael Cree will discuss our approach using 3D visual landmarks computed from sequences of camera views. The visual motion from a camera on a robot carries important cues as to the true heading of the robot and the estimation of depth to objects. The characteristics of an object's motion (eg a human) can be used to identify and track that object. John Perrone will discuss his work on self and object motion as it relates to the robot navigation and mapping problem.


A practical lock-free queue algorithm and its verification

Lindsay Groves
School of Mathematical and Computing Sciences, Victoria University of Wellington
Tuesday, 27 July 2004
(Joint work with Simon Doherty, Victoria University, and Mark Moir and Victor Luchangco, Sun Microsystems Laboratories, Boston.)

Lock-free implementations of concurrent data structures have been developed to overcome problems associated with the use of the traditional locking techniques, which render them impractical in large scale concurrent environments. Although they offer considerable performance improvements, lock-free algorithms are typically quite subtle and many of the published algorithms are incorrect - even ones accompanied by “proofs” of correctness. There is therefore a need to develop more rigorous techniques for verifying such algorithms.

This talk will describe one of the most widely used lock-free algorithms: Michael and Scott’s lock-free FIFO queue implementation, showing informally how it works. It will then present an overview of a semi-automated formal verification of a slightly optimised version of the algorithm. We verify the algorithm by proving a simulation relation between an automaton modelling the algorithm and one modelling the behaviour of a FIFO queue. An interesting aspect of the proof is that it requires a combination of two simulations: a forward simulation from the implementation automaton to an intermediate automaton, and a backward simulation from the intermediate automaton to the specification automaton. These automata are encoded in the input language of the PVS proof system, and the properties needed to show that the algorithm implements the specification are proved using PVS’s theorem prover.


Performing software

Professor Colin Beardon
Department of Computer Science, The University of Waikato, Hamilton
Thursday, 22 July 2004
Both the practical and the theoretical aspects of my work will be addressed in a presentation that will contain four short, interrelated parts each approaching the matter from a new direction. These will be: a practical demonstration of the 'Visual Assistant' software that I wrote for theatre set design; an account of how the software was designed in conjunction with theatre practitioners; a discussion of key concepts and terminology; and my underlying materialist philosophical stance related to the concept of the 'Digital Bauhaus'. We will then be clear to approach the main topic: how such components may be assembled to define a future research programme, facility and projects here at Waikato.


The WAND network research group in Waikato

Dr Murray Pearson and Dr Richard Nelson
Department of Computer Science, The University of Waikato, Hamilton
Tuesday, 20 July 2004
The WAND Group has two FRST research contracts: Connecting Remote Communities (CRCNet) and Measurement and Simulation.

Richard Nelson: The Measurement and Simulation project aims to provide an basis for integrating network performance measurement so as to make it accessible for enterprises without dedicated expertise. This is based on the previous WAND work in both these areas and includes topology discovery, trace capture, traffic modelling, network emulation and visualisation.

Murray Pearson: The CRCNet project has built several networks into rural and remote New Zealand using wireless technology. Murray will provide and introduction and status update on this project with a full presentation at a later seminar.


Investigations in multi-hop wireless ad hoc networks: TCP throughput and energy efficiency

Rajeev Shorey
IBM India Research Laboratory, New Delhi
Thursday, 15 July 2004
We study the performance metrics associated with TCP-regulated traffic in multi-hop, wireless networks that use a common physical channel (e.g., IEEE 802.11). In contrast to earlier analyses, we focus simultaneously on two key operating metrics the energy efficiency and the transport-layer (TCP) throughput. Using analysis and simulations, we show how these metrics are strongly influenced by the radio transmission range of individual nodes. Due to tradeoffs between the individual packet transmission energy and the likelihood of retransmissions, the total energy consumption is a convex function of the number of hops (and hence, of the transmission range). On the other hand, the throughput of a single TCP session decreases with a decrease in the transmission range. The overall achievable TCP throughput in an ad-hoc network thus involves a tradeoff between the reduced throughput of an individual flow and the greater degree of spatial reuse possible. As a consequence of this tradeoff, the overall network capacity turns out to be a concave function of the transmission range. We analyze how parameters such as the node density and the radio transmission range affect the overall network capacity under different operating conditions. Our analysis shows that capacity metrics at the TCP layer behave quite differently from the capacity results previously presented in literature. We then extend the work and examine the sensitivity of the TCP-layer capacity to the speed of the nodes and the number of TCP connections in an ad hoc network. By incorporating the notion of a minimal acceptable QoS metric (loss) for an individual session, we show why the QoS-compliant capacity is a more accurate metric for studying the capacity of TCP traffic in an ad hoc network. Finally, we study the dependence of capacity on the source application (Telnet or FTP traffic) and on the choice of the ad hoc routing protocol (AODV, DSR or DSDV). In the final part of the talk, I will discuss some of our recent work in multi-hop wireless ad hoc networks.


Snapshots of FM work at Waikato

Members of the Formal Methods Group
Department of Computer Science, The University of Waikato, Hamilton
Tuesday, 13 July 2004
In this seminar we will give snapshots of three, amongst several, of the areas of work in Formal Methods currently being undertaken in the Department.

David Streader and Steve Reeves:
Formal methods are best not seen and hence best embedded in a tool that relates what is wanted to what has been provided. In this context formal methods can be seen as essentially a translator between language of specification and an implementation language. To make such translations useful we need languages that are easy to use and understand. To this end we have been looking at combining different formal languages so as to get the best of both.

Robi Malik:
Finite-state machines are used to model the behaviour of software for controlling technical processes that are often safety-critical. Realistic systems can have very large state spaces, with 1020 and more states. We investigate new methods to explore the behaviour of large finite-state machines in order to find errors before software is actually used.

Mark Utting:
We will give an overview of the CZT (Community Z Tools) project, which is developing a suite of XML-based tools for the Z specification language, using Java, XML Schema, XSL, Velocity, Ant, Jaxb, Xerces, Xalan, JFlex, JavaCup, Bean Shell, etc. We will also discuss two applications of formal methods to WEKA.


Emotional design

Dr Donald A. Norman
Nielsen Norman Group and Northwestern University, Evanston, IL
Monday, 5 July 2004
Dr Norman brings a unique mix of the social sciences and engineering to bear on everyday products. Although he is a strong advocate of human-centred design and simplicity and perhaps best known for his book, "The Design of Everyday Things" he now wants to ensure that products appeal to the emotions as well as to reason.


Anti-spam measures and the digital divide: an exploratory study

Professor Christopher Lueg
Charles Darwin University, Australia
Thursday, 24 June 2004
Proliferation of unrestricted Internet access has brought the community unsolicited commercial email, better known as spam. Underestimated for quite some time, spam is now recognized as a problem costing the community billions of dollars per annum. One of the direct impacts of the spam flood is the widespread deployment of anti-spam measures, such as email filters and block lists. In this talk, I will summarize scholarly and anecdotal evidence suggesting that apart from reducing the spam load, anti-spam measures are also undermining the email system in terms of reliability and usability. Furthermore, I will discuss findings suggesting that anti-spam measures are also contributing to establishing a digital divide between those having a choice as to how they access email (both from a technical perspective and an educational point of view) and those who are not in this favorable position.


Spatial hypertext for DL users

Mr George Buchanan
Department of Computer Science, The University of Waikato, Hamilton
Tuesday, 22 June 2004
Spatial hypertext is primarily used for the task of "information structuring" - where a user organises documents as part of some information work. This may involve organising by author, topic, priority or any other characteristic the user perceives as important. Multiple organisations may be used at the same time - e.g. colour for priority, position for topic. Unlike traditional hyperlinked hypertexts (such as the Web), space and visual properties are used to articulate relationships, not 'point-and-click' navigation in a graphy of links.

Recently, spatial hypertext has started to be used with digital library systems to supplement the information seeking of readers in a library. With the combination of spatial hypertext and digital library, users can both retrieve documents and organise them for their own use - just as they would on the desk in a physical library.

This talk will present George’s work in this field, and give an overview of the motivations behind spatial hypertext and the reasons that researchers are now looking at combining the two systems.


Text summarization using lexical cohesion

Assistant Professor Yllias Chali
Department of Mathematics & Computer Science, University of Lethbridge, Alberta, Canada
Tuesday, 15 June 2004
With the continuing growth of the quantity of on-line text information, triggered in part by the growth of the World Wide Web, it is especially useful to have tools which can help users digest information content. Text summarization addresses this need by taking a source text, selecting the most important portions of it, and presenting coherent summary to the user in a manner sensitive to the user's or application's needs. The goal of this talk is to show how these objectives can be achieved through an efficient use of lexical cohesion. The current work addresses both generic and query-based summaries. I will present an approach for identifying the most important portions of the text which are topically best suited to represent the source text according to the author's views or in response to the user's interests. This identification must also take into consideration the degree of connectiveness among the chosen text portions so as to minimize the danger of producing summaries which contain poorly linked sentences. I will present a system that handles these objectives, discuss its performance, and compare it to other systems in the context of Document Understanding Conference (DUC) evaluation. Finally, I will outline some future works.


A strategic management framework for improving performance in a distributed manufacturing enterprise environment

Dr Chi-Yu Huang
Department of Computing and Mathematics, Manchester Metropolitan University, Manchester, UK
Tuesday, 8 June 2004
The growth in the development of distributed manufacturing systems necessitates increases in efficient technologies in order to facilitate management of the networked resources. Systems such as Manufacturing Automation Protocol (MAP) already exist specifically designed for managing networked manufacturing devices. However, MAP has been criticised for being over-complex and expensive for small-medium sized enterprises. The Simple Network Management Protocol (SNMP) was designed for managing TCP/IP networks, but it can also be used for managing networked manufacturing devices. SNMP, like MAP, is based upon a client/server architecture, and both have scalability issues. I discuss the use of mobile agent (MA) technology as means of managing a domain of manufacturing devices. MAs are not without their own scalability issues. A management framework is proposed that delivers improved performance using MA technology.


Machine Learning at Waikato

Dr Eibe Frank, Dr Mark Hall, Assist Prof Geoff Holmes, Dr Bernhard Pfahringer and Dr Tony Smith
Department of Computer Science, The University of Waikato, Hamilton
Tuesday, 1 June 2004
Machine learning at Waikato covers a broad range of topics and it is our aim in this seminar to briefly outline our interests. A very brief overview of the field will be followed by talks about the various projects that we have currently underway. These include the: FRST project, MILK (multi-instance learning kit), CLUSTERING, ILP (Inductive Logic Programming), TEXT CLASSIFICATION, KNOWLEDGE FLOW, TIME SERIES, BIOINFORMATICS and ML APPLICATIONS.


Realistic documents: A bizarre homage to an obsolete medium?

Jack Yi-Chun Chu, Dr David Bainbridge and Professor Ian H. Witten
Department of Computer Science, The University of Waikato, Hamilton
Monday, 31 May 2004
For many readers, handling a physical book is an enjoyably exquisite part of the information seeking process. Many physical characteristics of a book-its size, heft, the patina of use on its pages and so on-communicate ambient qualities of the document it represents. In contrast, the experience of accessing and exploring digital library documents is dull. The emphasis is utilitarian; technophile rather than bibliophile. We have extended the page-turning algorithm we reported at last year's JCDL into a scaleable, systematic approach that allows users to view and interact with realistic visualizations of any textual-based document in a Greenstone collection. Here, we further motivate the approach, illustrate the system in use, discuss the system architecture and present a user evaluation. Our work leads us to believe that far from being a whimsical gimmick, physical book models can usefully complement conventional document viewers and increase the perceived value of a digital library system.


Digital libraries meet libraries

Dr Alfredo Sanchez
Visiting Professor, Department of Computer Science, The University of Waikato, Hamilton
Tuesday, 18 May 2004
R&D programs in digital libraries (DL) are typically undertaken by computer science groups, whereas traditional library organizations try to assimilate recent developments to bring their services, collections and personnel up to date. The digital libraries program at Universidad de las Américas, Puebla (UDLA) is a joint initiative of the Libraries Division and a CS research center. This has generated interesting challenges and opportunities. Some of the projects under UDLA's DL program include visualization interfaces and collaboration environments. The rationale for these projects, implementation issues and deployment experiences, as well as opportunities for digital libraries in libraries, are the focus of this talk.


"The need for speed"

Mr Rob Akscyn
President, Knowledge Systems
Tuesday, 11 May 2004
As organizations and their members increase their knowledge workload, the efficiency of their tools becomes paramount. Unfortunately most modern 'productivity tools' remain woefully suboptimal and pay high opportunity costs in terms of personnel time. Simply put - they waste precious resources at virtually every click of the mouse.

Using KMS - a hypertext system for managing knowledge in organizations - the talk will illustrate how the hypertext paradigm can be used to streamline knowledge work, not only for accessing knowledge, but also as a medium for its creation and maintenance. Rob will also highlight lessons he has learned since he began developing and using hypertext technology in 1978.


Performance of the AT&T/Lucent Corporate Wide Area Network

Dr Alan Holt
Department of Computer Science, University of Waikato
Tuesday, 4 May 2004
There are four words used in the order "the network is slow" that I have come to dread. For a decade I worked on the AT&T/Lucent corporate wide area network. I've seen network problems caused by sharks, rats, Mickey Mouse, Osama Bin Larden, disgruntled employees from Belgium telecom and careless technicians. Hell I've pulled out the odd wrong cable myself from time to time. But the vast majority of network problems I've ever experienced are caused by everyday users just "using" the network. Problem is there's just seems to be too many of the them using it. During this period I learnt that Network performance is dependent upon many factors and is just as much a political, ethical and psychological issue as it is a technical one.


Hierarchical hidden Markov model for information extraction

Lin-Yi Chou
Department of Computer Science, University of Waikato
Tuesday, 27 April 2004
Hidden Markov models are proven to be capable of assigning semantic labels to tokens in a variety of input types. For example, natural language, DNA sequence and etc. Hierarchical hidden Markov models (HHMMs) are an extension of HMMS; one that provides a better fit to data that are hierarchical structured. The primary motivation is to enable better modelling of the general tasks. Do HHMMs have advantages over HMMs for information extraction tasks?

Lin-Yi will apply different tasks (i.e. reference tagging problem) to address the issue.


What is the fuss about Agile methods?

Dr Frank Maurer
Department of Computer Science, University of Calgary, Calgary, Canada
Tuesday, 20 April 2004
Agile software processes (like Extreme Programming, Scrum, and others) are a recent trend in the software industry. According to Giga Information Group Inc., more than two-thirds of all corporate IT organizations will use some form of agile software development process within the next 12 months. The presentation will give an introductory overview on agile methods. I will then discuss the benefits and issues with applying agile approaches and present some initial empirical evidence. Finally, I might present the MASE tool for supporting project planning and coordination of agile teams.


Databases, but not as we know them

Associate Professor Gill Dobbie
Department of Computer Science, University of Auckland
Tuesday, 6 April 2004
The Web has become a very useful tool to disseminate information to a diverse audience. Portals are springing up with all kinds of information, e.g. Rural News, Angling in Otago, and Typhoons. The information that is presented can either be gathered dynamically when requested or stored in some kind of repository. If it is stored in a repository, what kind of repository should be used? One possibility is a native XML database.

In this seminar we describe what a native XML database is, and investigate three research areas related to native XML databases. The seminar may raise more questions than it answers. This seminar assumes no prior knowledge of native XML databases.


Windows versus whiteboards: talking about Interfaces

Dr Michael Twidale
University of Illinois
Tuesday, 23 March 2004
Traditionally in HCI we think about designing an interface so that a (lone) user can interact (we might say 'talk') with the computer. Interface design then involves drawing pictures on the screen to make it easier for the user to manipulate and get information from abstractions - data types stored in the computer and manipulated by algorithms. We might consider in this case that the interface is a 'window' that allows us to look inside the computer at its data. What happens when we think of the interface less as a window and more as a whiteboard? That is, as a tool that we use while talking to another person in the room.

How can we design interfaces so it is easier to talk about them and their use to another person, especially when things go wrong?

How can we design interfaces that make it easier for one person to explain to another how to use them?

How and why might we want to design systems and interfaces that are less attention grabbing?

Mike shall illustrate this talk by examples from a project on ‘over the shoulder learning’, examining workplace peer help in the use of computer applications.


How to use the new library catalogue most effectively

Cheryl Ward
Subject Librarian, The University of Waikato
Tuesday, 16 March 2004
There is a new catalogue system at the University library. Cheryl is coming along to introduce us to the interface, search strategies and all. Come along and bring your questions, too.

She will mainly play by ear - so our questions do matter. She said, that when she gave the talk at the physics department, they kept on coming up with new questions for over an hour.


New paradigms in multimedia management and search

Dr Stefan Rüeger
Department of Computing, Imperial College London
Tuesday, 9 March 2004
The talk is about content-based multimedia retrieval, eg search of similar still images to given ones, finding a News story broadcast over TV by providing visual examples or finding a music piece by humming it. Our retrieval approach is not based on manual annotation but on /automated/ processing and feature extraction. Some of the challenges in this approach are given by (i) the semantic gap between what computers can index and high-level human concepts and (ii) the polysemy inherent in visual material and the corresponding wide range of information need by the user. We address these challenges with interactive retrieval methods, the use of relevance feedback - explicit or implicit and flexible learning algorithms that adapt themselves to the user and their information need.

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 providing examples of what is relevant. The talk will cover current work of our multimedia information retrieval team at Imperial College London:

  • News Search Engine and Summarisation
  • Search, Relevance Feedback and Browsing in Image and Video Databases
  • Information Navigation within a set of Documents


Research in computers & creativity

Professor Colin Beardon
Department of Computer Science, The University of Waikato
Tuesday, 24 February 2004
Until recently, research into creativity and computers usually originated from cognitive science and the models of human creativity developed by Margaret Boden and others. The past ten years has seen the increasingly sophisticated use of computers by artists, designers and others who have traditionally been thought of as being particularly creative in their own work. It has coincided with organisational changes within higher education that have led to the creative disciplines awarding research degrees and being assessed for their research. All of these factors have led to a major reassessment of research in areas concerned with creativity, often with wider implications.

This seminar will look at some of the emerging research laboratories involved with digital technologies and creative practices. It will also look at some of the values behind creative practices and why this sometimes creates friction with more traditional academic practices. It has also led to problems with traditional research granting or assessment agencies, and created interest in the notion of ‘research-equivalent’ activity.

The seminar will then consider our own situation where the department’s interest in this field has been considerably extended by the launching of the BCGD. The implications of this development, the potential for a new research group in ‘Computers & Creativity’ and some specific research projects will be outlined … still leaving plenty of time for discussion.


Events Index