CHAI Seminar Schedule
2008
Here is the schedule for the Computer Human Adapted Interaction Research Group seminars.
They range from formal presentations of mature work by members of the group or visitors to less
formal reports of work-in-progress. The regular meeting time is 11:00am Wednesday in the School of IT Building Seminar Room 123.
Subscribe to the CHAI Seminar Calendar
Weekly Seminars
Special Seminars
:: WWDC 2008 Trip Report
Speaker:
Greg Darke
:: Pervasive Personalisation of Location Information: Personalised Context Ontology
Speaker:
William Niu
Description:
There is considerable value in personalising information about people's location. Personalised Context Ontology (PECO) is an ontology for a building, and with PECO, we can provide personalised descriptions of the relevant people. For pragmatic reasons, it is important that PECO is created semi-automatically, making flexible use of a range of sources. For reasons of user control, it is important that PECO can be used to explain the personalisation. This paper describes PECO and how it is created for reasoning about a building. We also describe its use in an application called Locator, which presents information about the people in a building. PECO enables Locator to provide personalised information in two ways: it shows people of relevance and it makes use of personalised location labels. At the same time, PECO enables the user to scrutinise the reasoning about the personalisation. We report a study with eight users in which we compare a personalised and a non-adaptive versions of Locator. This indicates that people preferred the personalised version even though they could complete the designed tasks with both systems.
:: CHI 2008 Trip Report
Speaker:
Anthony Collins
Description:
Anthony will give a trip report of CHI 2008, the SIGCHI conference on human factors in computing systems.
:: Location Conflict Resolution with an Ontology
Speaker:
William Niu
Description:
Location modelling is central for many pervasive applications and is a key challenge in this area. One major difficulty in location modelling is due to the nature of evidence about a person's location; the evidence is commonly noisy, uncertain and conflicting. Ontological reasoning is intuitively appealing to help address this problem, as reflected in several previous proposals for its use.
This paper makes several important contributions to the exploration of the potential power of ontologies for improving reasoning about people's location from the available evidence. We describe ONCOR, our lightweight ontology framework: it has the notable and important property that it can be semi-automatically constructed, making new uses of it practical. This paper provides a comprehensive evaluation on how ontological reasoning can support location modelling: we introduce three algorithms for such reasoning and their evaluation based on a study of 8 people over 10-13 days. The results indicate the power of the approach, with mean error rates dropping from 55% with a naive algorithm to 16% with the best of the ontologically based algorithms. This work provides the first implementation of such an approach with a range of ontological reasoning approaches explored and evaluated.
:: Trip Report on CHI 2008 and the Collocated Social Practices Surrounding Photos Workshop
Speaker:
Trent Apted
Description:
Trent will give a trip report of CHI 2008 (the SIGCHI conference on human factors in computing systems), and the Collocated Social Practices Surrounding Photos workshop.
:: PIM 2008 Workshop Report
Speaker:
Anthony Collins
Description:
Anthony will give a report of the Personal Information Management (PIM) 2008 workshop, held this year at CHI in Florence, Italy.
:: Activity-based file systems
Speaker:
Greg Darke
Description:
This seminar is an overview of Greg Darke's honours thesis project. In this project Greg plans to investigate ways to link digital artefacts to higher level activities.
He will describe the ideas of activity theory which he uses as his grounding, as well as related systems that have already been built and his idea for a file-system monitor based implementation for the Apple Mac platform.
Speaker Bio:
Greg Darke is an honours student in the School of IT. He is currently in his 4th year of his B. IT degree.
:: Who am I: Files and their identities
Speaker:
Sam Thorogood
Description:
This seminar presents Sam's proposed thesis project investigating how to help the average computer user be more productive in managing their evolving file collection, across multiple machines and dealing with variants and versions. He will describe simple 'best-effort' operating system tools that help maintain the identity, uniqueness and freshness of documents and other files - not only over time, but also over multiple loosely-coupled systems.
Speaker Bio:
Sam Thorogood is an honours student in the school of IT, after completing his B Computer Science & Technology (Adv) in 2007.
:: Honours Thesis Talk
Speaker:
James Bunton
Description:
Personal information management is one of the most important aspects of computing today. People rely on their computers to store large amounts of their critical data. File system interaction is the way that users store and retrieve their documents from the operating system. This talk will give an informal review of previous work in this area, highlighting some possible areas of future research.
Speaker Bio:
James Bunton is an honours student in the School of IT. He is doing a Bachelor of IT, majoring in Principles of Computer Science and Software Engineering. Currently he is studying in the CHAI research group and doing research on novel ways of accessing files with rich metadata.
:: Automated feedback on the quality of an OO solution
Speaker:
Robert S. Rist (Faculty of IT, University of Technology, Sydney)
Description:
One of the hardest things to teach new programmers is object-oriented design. It is possible to provide explicit design criteria to evaluate the quality of an OO solution and to provide examples of good solutions, but many students cannot apply these lessons to their own code. Tutors and lab assistants can provide such feedback, but one-on-one personal contact time is severely limited and is often spent on basic programming, on getting the code to compile and run.
The problem with automating this task is variability: almost every student writes a different solution. This variability can be traced back to three sources. First, different names can be used for the classes, variables, and methods. Second, the code can be placed in different locations: different classes, different methods, and different orders and groupings of the code in a method. Third, different implementations of the same plan can be coded. The first source of variation can be overcome by a list of synonyms, supported by code matching from a model solution. The second source of variation can also be overcome by matching the code from a model solution onto the structure of the student solution. The third source is the most difficult to handle; it requires matching to a set of stored code patterns or schemas.
The talk will describe the automated feedback tool I am building while on sabbatical here in the School of IT. The input to the tool is a model solution, a set of synonyms, and an example solution to be evaluated. The tool will match the model solution to the example, provide feedback on how the two differ, and give hints about how to improve the example. It will provide explanations (in terms of design rule violations) to generalise the lesson (the design rule) from the presented example. It should allow automatic marking of an example solution, if the coverage is good enough. It may even turn out to be a powerful research tool that finds new programming patterns or schemas, the parts of the example solution that cannot be matched to the model.
Speaker Bio:
Robert Rist graduated in 1990 from Yale University with a doctorate in Cognitive Science; while there, he worked closely with the Soloway group. His thesis described and empirically supported a model of software design that included previous models(top-down, bottom-up, opportunistic) as special cases of a more general approach: schema construction and retrieval. Since that time, he has been a leading figure in the empirical study of programming, and has taught programming, AI, and systems modeling at the University of Technology, Sydney. He helped design the new IT degree at UTS in 2005, led the programming group that designed and implemented the new programming subjects in 2006, and taught the new first-year programming subjects in 2007.
:: Refaction Reflection
Speaker:
Matthew Wardrop
Description:
The Reflect project evolved over many years through an ad-hoc process involving many different developers. As a result, the code left a lot to be desired, and adding even a simple new feature was quite an arduous task! Thus, what began as a task to merely enhance the code autotesting backend (to add support for inserting data into the learner model), quickly ballooned into an effectively complete rewrite of the project. Whilst it is certainly not yet feature complete, I hope that with a fleeting ten to fifteen minute juxtaposition of the two versions, you will be as convinced as I am about the trend in progress; that is to say, an inarguably positive one.
Speaker Bio:
Matthew Wardrop is entering the second year of his BSc(Adv) undergraduate this year, with the hope of majoring in Physics and Mathematics. Since a very young age, he has enjoyed pulling apart and reconstructing his toys, and programming became a logical extension of that. Whilst heading, at this stage, toward a vocation in Theoretical Physics, Matthew cannot foresee ever losing his fascination with IT, or an end to his pursuits in programming. "Deriving algorithms," he says, "wheresoever they may be implemented, and at whatever scale, is just about as beautiful and intriguing as anything else I can imagine." He very much looks forward to sharing his ideas, and efforts, however paltry they may be by comparison to your own.
:: Peregrine: An Information Display Built on Locator
Speaker:
James Constable
Description:
The Locator system collects a wealth of information about users as they go about their day-to-day lives, and one of the many challenges involved with this system is finding useful ways of presenting the data.
This talk (the second part of a two-part talk) looks at Peregrine (PERsonalised Electronic GReeting and INformation Engine), an interface system that detects the presence of nearby users and provides them with personalised, context-sensitive information in a way that fits seamlessly into the surrounding environment and requires no direct user-interaction.
Speaker Bio:
James Constable is a summer vacation scholar with the CHAI research group. He is about to commence his third year of a Bachelor of Information Technology.
:: Student inference with Reflect
Speaker:
Steve Bian
Description:
One of Reflect's goals is to provide a means of inferring student ability from their performance at Reflect tasks, and provide insight into their mastery of concepts being taught. To this end, I was set the task of investigating applying Item Response Theory (IRT) and Bayesian Networks (BN) to existing Reflect data and judge how effective they are for our needs.
My presentation will mainly focus on IRT, how it was applied to Reflect data from 2007, how well it estimated each student's ability, and also how its ability estimates can be used as input into a BN representing a course such as SOFT2130.
Speaker Bio:
Steve is a undergraduate student studying BE/BSc (electrical engineering and physics). He has been a tutor for SOFT2130, and has helped run the ACM training sessions and competition in 2007. He likes to read, builds robots on weekends, and annoy industrial reps.
:: Reflect - Learning Concepts and their Ontology
Speaker:
Christopher James Ackad
Description:
Reflect is an online submission/assessment system aimed at encouraging reflective learning in students. At the moment this system is implemented in a second year C programming course with plans to expand into other courses. To assess students' knowledge one needs to define what is to be taught. The learning concepts in Reflect define a construct of the course's (or instance's) body of knowledge.
The focus of the talk will be on how learning concepts have been enhanced in Reflect via the implementation of an ontological structure that provides greater depth and the ability to exchange the learner models between instances of Reflect as well as other applications. It will discuss a sample implementation of an author and student user interface that allows the creation and easy viewing of an ontology of learning concepts in Reflect. It will also explore topic maps as a viable transfer medium and the opportunity of implementing a database version of Reflect.
Speaker Bio:
Christopher Ackad is a summer vacation scholar with the CHAI research group. He is about to commence his third year of a Bachelor of Computer Science and Technology.
:: LAVAlamp and Peregrine: Two Information Displays Built on Locator
Speaker:
James Constable
Description:
The Locator system collects a wealth of information about users as they go about their day-to-day lives, and one of the many challenges involved with this system is finding useful ways of presenting the data. This talk looks at two different approaches to the problem, both aimed at creating attractive yet useful interfaces that fit seamlessly into the surrounding environment and require no direct user-interaction.
The first part of the talk will be focussed on a user evaluation of the LAVAlamp system (developed by Raymes Khoury). The second part of the talk will look at Peregrine (PERsonalised Electronic GReeting and INformation Engine) - an interface system that detects the presence of nearby users and provides them with personalised, context-sensitive information.
Speaker Bio:
James Constable is a summer vacation scholar with the CHAI research group. He is about to commence his third year of a Bachelor of Information Technology.
:: Adding additional sources of evidence to Locator
Speaker:
Glen Pink
Description:
The Locator system currently uses a variety of sensors to provide evidence as to where an individual is located - these sensors detect either the presence of Bluetooth devices or a user's activity at a computer. This talk looks at additional sensors added over the course of my summer scholarship, considering the use of Nike shoe transmitters, WiFi, RFID cards, infra-red sensors, and door switches. These new sensors are intended to exist alongside the existing sensors as unobtrusive sources of evidence, as well as increase the accuracy of location via the use of custom resolvers.
Speaker Bio:
Glen Pink is a summer vacation scholar with the CHAI research group. He about to commence his third year of a Bachelor of Information Technology.
:: Teamwork Pattern Finder: plug-in for Trac to monitor group behaviour
Speaker:
Vincent Daviet
Description:
Vincent will present the plug-in he built for Trac, a software project management tool for teamwork. Students of the School of IT also use Trac to help them manage their project. This tool pre-processes the huge amount of students' data, using different Data Mining techniques. Teachers can add and save the patterns that they think are indicative of "good" or a "poor" behaviour. Then, the plug-in is able to search and to compare the results with the interesting patterns found by teachers. These information are displayed to give a proactive feedback to the learners about their teamwork.
Speaker Bio:
Vincent Daviet has done a 6-month internship at the School of IT, University of Sydney. He is an Engineering Computer Science student at the UTBM (University of Technology Belfort-Montbeliard, France). Prior to this, he completed a 2-year Technical degree in Networking and Telecommunication (IUT - GTR, Annecy, France).
:: Narcissus: an Interactive Visualisation for Mirroring Activities of Small Long-Term Groups
Speaker:
Kimberley Upton
Description:
Working effectively in groups is hard, particularly for projects that span some months. Groups can be derailed by a range of problems including poor leadership, poor communication, failure to monitor progress, and social loafing. Learning to work effectively in a group is a difficult but necessary measure to ensure group success.
Groups using electronic support tools provide huge amounts of trace data, which has the potential to provide the group with insights into the effectiveness of their own operation. The Narcissus project exploits this data to support the groups. The approach is to use an interactive visualisation to mirror the activities of the group members, which gives them the opportunity to see the dynamics and progress of their group.
This talk outlines the design and evaluation of the initial protoype, and discusses the upcoming deployment of a plugin for Trac and future directions for the project.
Speaker Bio:
Kimberley Upton is a recent Honours student, having completed a Bachelor of Computer Science and Technology (Advanced) at the University of Sydney. She is doing a Summer project with the Computer Human Adapted Interaction Research Group, supervised by Judy Kay.
:: Data Mining Learning Traces from Trac (Part II)
Speaker:
Dilhan Perera
Description:
Students in the SOFT3300 course work in groups of 5-7 to develop a substantial software solution for a real client. These groups use the online project management and collaboration tool TRAC to aid communication and management during development. The aim of this project has been to use various data mining techniques to extract useful information from the data left behind as these students use TRAC. During this talk I will present the work I have during my summer scholarship, which represents an extension of the previous work started in 2006. This includes a comparison of results from the 2007 cohort with the 2006 cohort, as well as an identification of the limitations of the current techniques and an exploration of ways to overcome these limitations.
Speaker Bio:
Dilhan Perera is a summer vacation scholar with the CHAI research group at the School of IT. He is about to commence his fourth year in the combined Bachelor of Science / Bacherlor of Commerce course at the University of Sydney.
:: Evaluating tabletop file and personal information management
Speaker:
Anthony Collins
Description:
File system and personal information management is a task that is typically performed with a conventional personal computer, by a single user. Digital files and personal information (for example, e-mail and calendar appointments) are used daily to conduct tasks for a variety of purposes, from personal organisation, to social communication, and to corporate planning. Moving away from the desktop paradigm and its primarily single-user interaction methods allows new possibilities for both sharing personal information, and collaborating on the management process. The interactive tabletop -- a novel medium that has recently attracted significant research interest -- supports collocated collaboration with a large, shared workspace, where people can sit face-to-face and interact with it simultaneously. While users may not perceive a tabletop interface as a conventional personal computer, a form of file system is important for supporting collaborative tasks at the tabletop. Hierarchical navigation of file systems has become the standard in conventional personal computers, although the properties and constraints of tabletop interfaces call for rethinking standard approaches to file system interaction.
In this talk, I will first present our past work on evaluating a novel tabletop file system interface called Focus. Second, I will outline the proposed design for our next major study of tabletop file access (designed to build on our previous studies), which is being conducted in conjunction with the HxI Initiative Braccetto project. The study will involve a qualitative comparison of the Focus interface with Windows Explorer (a legacy file system interface designed for single-user interaction) to assess how effectively these systems support collocated collaboration at an interactive tabletop.
Speaker Bio:
Anthony Collins is a PhD student with the Computer Human Adapted Interaction (CHAI) Research Group at the University of Sydney. Prior to this, he completed a Bachelor of Computer Science and Technology (Advanced) degree (also at the University of Sydney), and graduated with first class honours.
:: Personal Metasearch
Speaker:
Paul Thomas (CSIRO ICT Centre)
Description:
Future search tools will have to work with the tremendous range of online information available to users: including the entire Web but also corporate sources such as subscription services or databases and personal sources such as calendars or email archives. This diversity of sources suggests a novel model of personalisation, where data -- and not necessarily algorithms -- vary considerably from person to person.
This talk motivates and introduces "personal metasearch", our response to this search problem, and considers problems of building and evaluating a personal metasearch engine.
Speaker Bio:
Paul is a postdoctoral fellow at the CSIRO ICT Centre, working largely on problems in distributed and personal information retrieval. He completed his PhD, at the ANU, earlier this year.
:: SenseCam, Somniloquy and Force Sensing: An overview of research projects in the Sensors and Devices group at Microsoft Research Cambridge
Speaker:
James Scott
Description:
By prototyping new hardware, we enable research into a number of areas of computing affecting end users. I present three examples from different threads of research in Microsoft Research Cambridge's Sensors and Devices group, in wearable sensors, energy-aware computing and for mobile device interaction.
SenseCam is a wearable camera which has shown huge potential in aiding memory loss sufferers (e.g. with Alzheimer's) recall events they would otherwise forget. Somniloquy is a modest hardware/software addition to PCs allowing them to respond to network events by "talking in their sleep". Force Sensing is a new interaction technique for mobile devices using twisting and bending forces to provide button-free interaction for large-screen devices.
Speaker Bio:
James Scott is a Researcher in the Sensors and Devices Group, which is part of the larger Computer Mediated Living Group at Microsoft Research Cambridge in the UK. He joined Microsoft in January 2007. Prior to this, he spent four years at Intel Research, which he joined after finishing his PhD at Cambridge University in 2002. He received his BA in Computer Science in 1998, also from Cambridge University.
His research spans a broad set of topics in ubiquitous computing, including mobile device hardware, sensors and interaction techniques, wireless networking, power-aware computing, location sensing, and privacy/security issues.
:: Interacting in immersive environments and understanding user's experience
Speaker:
Leila Alem and Frederic Merienne
Speaker Bio:
Prof Merienne is head of the Imaging Institute of the LE2I (Laboratoire Electronique, Informatique et Image) Laboratory of ENSAM, France. His group in France are researching a number of applications of Virtual Environments, including applications in the medical, aviation, architectural, cultural and computer-aided design fields. He is currently spending a year in the Networking Technologies Laboratory of the CSIRO ICT Centre, where he is extending the capabilities of his technology taking advantage of CSIRO's expertise in telecollaboration and Human Factors.
Dr Leila Alem is a senior research scientist at CSIRO ICT Centre. Her group is conducting research is Human Factors in tele collaborative settings. Leila is interested in formalising and measuring user's experience when engaged in synchronous computer mediated collaboration, more specifically users' sense of presence. In recent years she has conducted several laboratory and field experiments to investigate the various factors affecting users' sense of presence. She has also investigated novel approaches for developping objective measures of presence using gaze tacking and analysis methods as well as using linguistic analysis approaches.
:: System Support for Pervasive Computing
Speaker:
Prof. Christian Becker
Description:
In the vision of Pervasive Computing computers pervade our daily environment - mostly as embedded systems that augment our surrounding. Applications can utilize a number of services that are available in the physical proximity in order to offer their users services tailored to their current context. Due to mobility and effects, such as power saving, services will fluctuate. Applications have to adapt to compensate fluctuations as well as make use of "better" services that become available.
In the Peer to Peer Pervasive Computing project 3PC we have investigated middleware concepts that allow to establish a so called smart peer group that is formed from devices with a common mobility pattern. Within this smart peer group, resources are shared. We named this middleware BASE and used it as a technical foundation to investigate concepts and mechanism for automated application adaptation. In our component system PCOM contracts between components are enriched in their semantics and thus allow the system to automatically substitute a component in case of fluctuations or availability of a better candidate.
The talk closes with an outlook to current research that builds on the concepts developed in 3PC.
Speaker Bio:
Christian Becker is a full professor for Information Systems at the University of Mannheim since 2006. Prior to this he was a visiting professor for distributed systems at the University of Duisburg-Essen in Spring Term 2006. He studied Computer Science at the Universities of Karlsruhe and Kaiserslautern where he received the Diploma in 1996. From 1997 till 2001 he was a researcher at the distributed systems and operating systems group at the University of Frankfurt where he received his PhD in 2001 with a thesis about Quality of Service Management in Distributed Object Systems. In 2001 he joined the distributed systems group at the University of Stuttgart as Post Doc. His research focussed on system support for Pervasive Computing and Context-Aware Computing. He was a member of the Nexus project that investigates concepts for global scale context management. In 2004 he received the venia legendi (Habilitation) for Computer Science (Informatik). Christian's research interests are distributed systems and Context-Aware Computing.
Past Seminars
| 
![[Cellerator]](Images/cellerator.png "Cellerator Home"){kind=small}