|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
2002 Seminar SeriesSee below for detailed abstracts
Abstracts High Performance Computing on-board Small Satellites Abstract: Small remote sensing satellites take pictures for environmental research, e.g. looking f or forest fires, for oil spills, observing vegetation and settlements and potentially givin g storm warnings. The amount of pictures that can be taken exceeds the number of pictures that can be downloaded to the ground station by orders of magnitude. Thus the aim of on- board image processing is to sort out images that do not contain useful information, to segment images according to usefulness and to compress those parts that seem to be of little or no value to the users. State-of-the-art on-board computing is (surprisingly) characterized by the use of very slow, very expensive but space hardened (i.e. resistant to high doses of radiation) processing elements. In this talk a radically different approach is presented: We aim at using off-the-shelf, low-price, and high-performance components and connect them via a space-hardened fault-tolerant network into a parallel image processing engine. Even with this compute power real-time image analysis is a challenging task that requires redevelopment of the corresponding algorithms, some algorithms designed for this purpose are also presented. About the speaker: Professor Heiko Schroder is the Head of the School of Computer Science and Information Technology at RMIT.
An Introduction to Bioinformatics: Why and how we access genomic data As part of our post-graduate Master program, the Department of Biotechnology and Environmental Biology at RMIT has instituted a course in introductory Bioinformatics. Th is course is designed to facilitate the access to and use of databases and programs for the analysis of genomic and proteomics information. In this seminar I will summarise the con tent of this course. The need for bioinformatics will be discussed, and I will demonstrate some of the ``front pages'' on the WWW that permit access to the data. This will include the screening of databases for matching sequences (both DNA and protein), the alignment of multiple sequences and subsequent phylogenetic analyses, and the identification of protein motifs and the building of molecular models. About the speaker: Dr. Peter Smooker is a Lecturer in Bioinformatics and Immunology in the
Department of Biotechnology and Environmental Biology at RMIT.
Current Issues in Internet Security We live in an era that Eric Cole calls the "Golden Age of Internet Hacking." The lack of a robust security model built into Internet protocols from the start have created a network environment that's fraught with peril. In this talk, we will cover the basics of Internet vulnerabilities and hacker exploits, and both describe and illustrate the typical hacker's modus operandi: reconnaissance, exploitation (gaining access, elevating privileges, DOS and DDOS), control, maintaining access (back doors and trojans), and concealment. About the speaker: Dr. Berghel is currently Professor and Chair of Computer Science at the
University of Nevada at Las Vegas. He has held a variety of research and
administrative positions in industry and academia during his twenty-five
year career in computing. His current research focuses on Internet and
Web technologies, interactive and participatory computing environments,
including the design of virtual communities, and electronic information
management. His research work appears frequently in a variety of
scientific and technical venues, and his columns, editorials, and articles
appear regularly in such publications as Computer, the Communications
of the ACM, and Networker.
See also http://www.acm.org/top/people/berghel.html
Distributed Robotic Systems for Search and Rescue An application of major interest to roboticists is search and rescue.Robot searchers are expendable, cheap compared to human searchers and can enterareas that humans or dogs cannot. Interest in this application led to the foundation of the Robocup Rescue Robot League, a simulated search for people trapped within a collapsed building. To function in such an environment, the robot systems used must be physically and functionally robust. The inherent redundancy and parallelism of distributed robot systems makes them highly suited to search and rescue. However, developing distributed systems to effectively perform a search and rescue task is still difficult. Approaches to the design of distributed robotics systems include the use of team learning, and the use of concepts found in distributed biological systems. Of particular interest is the concept of emergence - the development of complex group behaviour through the continual interaction of simple agents. In my presentation, I will give an overview of the domain of robot search-and-rescue, and detail the particular aspects of robot team design I am investigating for my PhD. About the speaker:
Toby Richer graduated from Adelaide University with a Bachelor of
Science (Maths and Computer Science) and a Bachelor of Engineering
(Computer Systems)(1st Class Honours) in 2000. He is currently a PhD
student in the Intelligent Systems Group at the Advanced Computing
Research Centre, University of South Australia. His main interests
are Autonomous Robotics, Multi-Agent Systems and ALife.
Declarative and Procedural Goals in Intelligent Agent Systems An important concept for intelligent agent systems is goals. Goals have two aspects: declarative (a description of the state sought), and procedural (a set of plans for achieving the goal). A declarative view of goals is necessary in order to reason about important properties of goals, while a procedural view of goals is necessary to ensure that goals can be achieved efficiently in dynamic environments. In this paper we propose a framework for goals which integrates both views. We discuss the requisite properties of goals and the link between the declarative and procedural aspects, then derive a formal semantics which has these properties. We present a high-level plan notation with goals and give its formal semantics. We then show how the use of declarative information permits reasoning (such as the detection and resolution of conflicts ) to be performed on goals. About the speaker:
John Thangarajah is a PhD student in the School of Computer Science
and Information Technology at RMIT.
Efficient String Sorting Sorting of large numbers of values is one of the oldest problems in computer science, yet improvements continue to be made. Although many general-purpose sorting algorithms are widely applied to values of any type, the characteristics of strings mean than special-purpose algorithms can be considerably more efficient. During the 1990s, several string sorting algorithms were developed, the best of which require only one comparison per distinguishing character - the theoretical minimum. In this talk I discuss work in progress on string sorting (with PhD student Ranjan Sinha). While our methods require no fewer comparisons than earlier methods, we have found that efficiency gains can be made through use of randomisation, as demonstrated by experimental results on large sets of strings. Our results also suggest that significant further improvements may be available. About the speaker: Ass. Prof. Justin Zobel is an Associate Professor in the
School of Computer Science and Information Technology at RMIT.
Classification by Aggregating Emerging Patterns Data Mining aims at the discovery of useful information from large collections of data. Different kinds of knowledge can be mined, including association rules, characteristic rules and classification rules, to name a few. Emerging pattern (EP) is a recently proposed knowledge pattern which is combination of attribute-values to capture changes or differences of a target database w.r.t. a background database. In this talk, I discuss classification approaches based on new ideas using EPs. Experiments show that classifiers built with EPs have good predictive accuracy and usually outperform state-of-the-art classification systems. About the speaker: Xiuzhen (Jenny) Zhang is a lecturer at School of CS & IT, RMIT University.
Her research interests include
data mining and data warehousing.
The Eager Bidder Problem: A Fundamental Problem of Distributed Artificial Intelligence and Selected Solutions The contract net protocol is a widely used protocol in DAI, as it proved to be a flexible and low communication interaction protocol for task assignment. It is however not clear in which manner agents participating in a contract net should allocate their resources if a large number of contract net protocols is performed concurrently. If the agent allocates too many resources too early, e.g. when making a bid, it may not get any bid accepted and resources have been allocated while other negotiations have come to an end and it is no longer able to make bids for them. If it allocates resources too late, e.g. after being awarded the contract, it may have made bids for more tasks than its resources allow for, possibly all being accepted and resulting in committments that cannot be kept. We call this dilemma the Eager Bidder Problem. In this paper, we present an ad hoc solution and two more complex strategies for solving this problem. Furthermore, we introduce a new method based on a statistical approach. We describe these mechanisms and how they deal with the concept of commitment at different levels. There is no optimal solution for every problem setting, but each has advantages and disadvantages. Our discussion concludes with criteria for the decision which of these mechanisms is best selected for a given problem domain. About the speaker:
Michael Schillo received an MSc in natural language processing from
the University of Leeds, UK and holds a Diploma in informatics from
Saarland University, Germany. He is currently affiliated with the
German Research Center for AI (DFKI), where he is pursuing a PhD on
robustness in multiagent systems.
Organisational Web Search One organisational web may be very different from another but they all differ substantially both from the Internet and from the type of retrieval modelled by TREC ad hoc. A set of organisational webs have been chosen for study and may be characterised by extent, by the types of query submitted, and by the availability and usefulness of types of retrieval evidence such as title, descriptive metadata, and link information. A high proportion of queries submitted to organisational search engines seem to be attempts to name websites corresponding to entities such as a person, an office, a department, a product or a service. The effectiveness of various methods for satisfying such queries is evaluated across the organisations. About the speaker:
David Hawking is a Principal Research Scientist at CSIRO and leads the
team responsible for the Panoptic(TM) enterprise search engine. His
PhD topic was "Text retrieval over distributed collections". He has
coordinated the TREC VLC and Web tracks since their inception and,
with colleagues, has been responsible for the creation and
distribution of several widely used Web test collections. David is
Web Area Coordinator for the ACM SIGIR conference and a
Program Co-chair for SIGIR 2003. His interests encompass distributed
information retrieval, search taxonomy, search evaluation methodology, and all
aspects of Web search.
Teaching Computer Science in the 21st Century Many issues about university teaching are centred on what happens in a lecture theatre. However, it can be argued, for certain subjects at least, that tutorials have a far greater influence on students and their results. In this seminar, an approach to tutorials (and related assessment methods) based on aspects of problem-based learning will be discussed. Teaching Computer Science is an ongoing challenge, particularly when subjects can be delivered either partially or totally online. In this seminar some of the lessons learned from the experience of teaching Computing Theory will be discussed, with a particular focus on the way in which tutorials are run. Please note that this is intended as a colloquium, not a lecture; hence after some initial remarks, audience participation will be not just invited but actively sought. About the speaker:
James Harland is a Senior Lecturer in the School of Computer Science
and Information Technology at RMIT.
Debugging Multi-agent Systems Using Design Artifacts: The Case of Interaction Protocols Debugging multi-agent systems (which are concurrent, distributed, and consist of complex components) is difficult, yet crucial. We propose that the debugging process can be improved by following an agent oriented design methdology, and then using the design artifacts in the debugging phase. We present an example of this scheme which uses interaction protocols to debug agent interaction. Interaction protocols are specified using AUML and are translated to Petri nets. The debugger uses the Petri nets to monitor conversations and to provide precise and informative error messages when protocols aren't correctly followed by the agents. About the speaker:
David Poutakidis is a PhD student studying at the School of CS & IT at
RMIT. He is working on debugging multi agent BDI systems.
Building an Intelligent World: The State of the art in Distributed AI Research into embodied Artificial Intelligence is currently working on building humanoid robots whose perceptual, cognitive and motor faculties match those of human beings. Researchers in distributed AI on the other hand are working towards building a world comprised of individual agents (software robots) where no one particular agent is omnipotent but the system as a whole is intelligent. These agents are separate atomic entities that communicate and coordinate with each other to achieve goals that are beyond the capabilities of any one agent. The development of such an environment requires designing agents that can translate high-level objectives from human users into actions that achieve goals such as planning holiday trips, managing stock portfolios, and planning and executing disaster relief campaigns. In carrying out these actions agents may be required to communicate with each other, access databases for information, make decisions based on incomplete knowledge, generate plans in uncertain ever-changing environments and work in teams. This talk will look at the various technologies in academia and industry that are currently coming together to create a future where intelligence is ubiquitous and agents are an integral part of our everyday lives. A future, where intelligent machines will help us shop, pay our bills, filter our mail, screen our calls, answer our questions and generally carry out the mundane tasks we wish we didn’t have to do. About the speaker:
Rahul Singh is a Researcher with the Intelligent Software Agents Laboratory and a
graduate student at the Robotics Institute at Carnegie Mellon University. He received his
B.E. in Computer Systems Engineering from the University of South Australia, Adelaide
in 2000 and is currently pursuing a Masters degree in Robotics and Artificial Intelligence.
His research interests include Knowledge Representation, Reasoning and Planning.
Contact him at
http://www.cs.cmu.edu/~kingtiny
Parameter Control within a Co-operative Co-evolutionary Genetic Algorithm Typically GAs have a number of fixed control parameters which have a significant effect upon the performance of the search. This paper addresses the effects of self-adapting control parameters and the adaption of population size within the sub-populations of a co-evolutionary model. We address the need to investigate the potential of these adaptive techniques within a co-evolutionary GA, and propose a number of model variants implementing adaption. These models were tested on some well known function optimisation problems. The experimental results show that one or more of the model variants yield improvements over the baseline co-evolutionary model. About the speaker:
Xiaodong Li is a Lecturer in the School of Computer Science and Information Technology at RMIT University.
Retrieval, watermarking and view synthesis Abstract: In this talk I will give an overview of my research activities in multimedia compression and analysis. I will in particular concentrate on 3 projects.
About the speaker:
Nishan Canagarajah is now a Reader in Signal Processing at Bristol. He
joined Bristol in 1994 as a lecturer, promoted to a Senior Lectureship
in 1999 and a Readership in 2001. He has BA (Hons) and a PhD both from
the University of Cambridge, UK. His research interests include
multimedia signal processing, image and video coding, video
watermarking, stereoscopic video, imaged fusion, view synthesis and the
application of signal processing to audio and medical electronics. He
is widely supported in these areas by industry (TI, Sony, BBC, BAe,
Orange) EU and the EPSRC. He is currently managing grants worth £2M and
his research team include 12 Phd's and 8 post-doctoral researchers. He
has published more than 200 papers and two books. He is technical
Consultant of a company, that was a spin-out of his research in this
area. He is a committee member of the UK Visual Information Engineering
Professional Network.
Intelligent Decision Support for Industrial Applications Large-scale real world problems are complex, requiring smart ways to handle. Although there are many theories/methodologies that have been developed to address various aspects of real world problems, few of them have been found successful in solving these problems alone. The urge of innovations to address large-scale real world problems demands a multidisciplinary approach and requires cross-fertilization of disciplines, tailored problem-specific solutions, and a client-oriented attitude. In this talk, we will report on several recent industry/government funded research projects in intelligent decision support systems, including an Internet weed control advisory system, a sugar cane transport scheduler and a sugar mill pan stage supervisory control system. Several Intelligent Systems techniques will be described for dealing with various modeling and optimization tasks, e.g. constraint programming for scheduling, prediction model for spatial data sets, and expert knowledge acquisition and modeling. A meta modeling framework based on fuzzy logic will be presented for handling information and decision-making flows. About the speaker:
Xinghuo Yu received BEng and MEng from the University of Science and
Technology of China in 1982 and 1984 respectively, and PhD degree from
South-East University, China in 1988. He has been with School of
Electrical and Computer Engineering, RMIT University since March 2002,
where he is a Professor, Deputy Head of School, Director for Software
and Networks. He was with Central Queensland University between 1991
to 2002, where, before joining RMIT, he was Professor of Intelligent
Systems, Associate Dean (Research) of Faculty of Informatics and
Communication. He also held Visiting Professor positions at City
University of Hong Kong and Bogazici University (Turkey). His research
interests include control systems, chaos and anti-chaos, neural
networks, fuzzy systems, evolutionary computation, data mining,
intelligent agents and applications. Prof Yu has co-edited 5 research
books and published over 200 refereed papers in technical journals,
books and conference proceedings. He is serving as an Associate
Editor of IEEE Transactions on Circuits and Systems: Part I, and is on
the Editorial Board of International Journal of Applied Mathematics
and Computer Science. He is also on the Steering Committee of IEEE
Transactions on Mobile Computing. Prof Yu has recently been conferred
Emeritus Professor and Honorary Professor of Central Queensland
University. He is a Senior Member of IEEE, and was the sole recipient
of the 1995 Central Queensland University Vice Chancellor's Award for
Research. He also holds Guest Professor Positions in several major
Chinese Universities.
Content-based Image Retrieval
Saied will give an overview of his PhD thesis on content-based image retrieval (CBIR), focussing on the effectiveness of multiple-example queries. About the speaker:
Saied Tahaghoghi recently submitted his PhD in the School of Computer
Science and Information Technology at RMIT.
Tutorials, Assessment and all that jazz ... Many issues about university teaching are centred on what happens in a lecture theatre. However, it can be argued, for certain subjects at least, that tutorials have a far greater influence on students and their results. In this seminar, an approach to tutorials (and related assessment methods) based on aspects of problem-based learning will be discussed. Please note that this is intended as a colloquium, not a lecture; hence after some initial remarks, audience participation will be not just invited but actively sought. About the speaker:
James Harland is a Senior Lecturer in the School of Computer Science
and Information Technology at RMIT.
Interactive Role-Playing Multiagents for Decision-Making in Command and Control Multiagent teamwork has been used in military operations research to simulate autonomous entity reasoning in large-scale combat mission scenarios. We examine teamwork in desktop distributed command and control systems for monitoring, surveillance, mission planning and rehearsal in deployed operations. Our distributed multiagent mission planning system, HELO, uses agents to improve decision making and decision support for human operators involved in the collaborative mission planning task. HELO improves decision support by providing agent enabled simulation scenarios, and service providing agents that recognise the goals of their operator, helping achieve tasks by responding to queries, providing suggestions, automating activities, and disseminating information. HELO investigates teamwork modelling of collaborative mission planning where particular team structures and specific roles represent military personnel, simulation entities and service providers. Users can interact with agents by performing manipulation actions, assigning attributes, requests, intelligence and characteristics to agents, simultaneously throughout the distributed system, as a function of time. These agent profile manipulations affect the dynamic team formations used to model mission planners. The team structures may be added or compared against existing team formation relational structures using the CLARET pattern matching algorithm. Evaluations of these provide recognition of mission planning user requirements, influencing the degree of HELO's decision making assistance. These ideas are demonstrated through the current multi-agent language JACK and its JACK Teams extension. About the speaker:
Susannah Soon is a PhD candidate in the Department of Computer Science
and Software Engineering, The University of Melbourne.
The Generalised Shortest Path Generalized networks specify two parameters for each arc, a cost c(a) and a gain g(a). If x units enter an arc a, then x*g(a) exit. Arcs may generate or consume flow, i.e., they are gainy or lossy. Costs are charged per unit. The generalized shortest path problem is known as the restricted generalized uncapacitated transshipment problem and has been used for generalized net work) flow problems. It occurs e.g. in communication networks, where the gain expresses an increase of the transmitted amount of data, or in financial transactions with losses due to a decreasing stocks, money exchange fees or brokerage. The objective is finding the cheapest path for a unit flow from the source or to the target. The cheapest path is a simple path or it ends in a lossy cycle where all flow is consumed. We solve the generalized shortest path problems by an extension of the Be llman-Ford algorithm. The key is running the algorithm backwards. It runs in O(nmlogn), which improves to O(nm) in lossless resp. profitless networks and to O(nlogn+m) with non-decreasing costs and gains. The algorithm is simple r and at least a factor of n faster than the previous algorithms using linear programming) or complex parametric search and scaling techniques. To the contrary, the single-pair generalized shortest path problem is NP-hard, even with nonnegative costs and lossy resp. gainy arcs and uniform gains. Here, cycles are excluded. About the speaker:
Franz Brandenburg is a Professor in the Department of Mathematics and
Computer Science at the University of Passau in Germany.
Topics in Information Retrieval Part 1: Query Association for Effective Retrieval In the first half of my seminar, I will present joint work with Falk Scholer that will be presented at the 2002 Conference on Information and Knowledge Management. In this work, we introduce a novel technique for document summarisation in search engines which we call query association. Query association is based on the notion that a query that is highly similar to a document is a good descriptor of that document. For example, the user query ``richmond football club'' is likely to be a good summary of the content of a document that is ranked highly in response to the query. We show that associated queries are an excellent technique for describing a document: for relevance judgement, associated queries are as effective as a simple online query-biased summarisation technique. Part 2: Simple and Accurate Feature Selection for Hierarchical Categorisation In the second half of my seminar, I will present joint work with Wahyu Wibowo that will be presented at the 2002 ACM Symposium on Document Engineering. Categorisation of digital documents is useful for organisation and retrieval. While document categories can be a set of unstructured category labels, some document categories are hierarchically structured such as the Web hierarchies at Yahoo! and DMOZ. This talk discusses automatic hierarchical categorisation and, specifically, the role of words in the development of more effective hierarchial categorisers. Overall, we show that by using a few words from new documents, categorisation accuracy can be improved substantially. About the speaker:
Hugh Williams is a senior lecturer in the School of Computer Science and IT
at RMIT University in Melbourne, where has taught since 1994.
Computing the noncomputable: a quantum mechanical approach to computing the Turing halting function. Inspired by the concept of Quantum Computation, we re-examine the notion of computability, which holds a central position in Mathematics and Theoretical Computer Science. Up until now, computability has been delimited by the Church-Turing thesis, which stipulates that all the functions that are effectively computable are in fact Turing computable, and vice versa---thus restricting the intuitive and informal notion of computability to the well-defined mechanical operations of Turing machines. In particular, we consider the Hilbert's tenth problem: Given any polynomial equation with any number of unknowns and with integer coefficients (that is, any Diophantine equation): To devise a universal process according to which it can be determined by a finite number of operations whether the equation has integer solutions. This problem can be shown to be equivalent to the Turing halting problem. Both can be subsequently shown, with the well-known Cantor s diagonal arguments, to be Turing noncomputable. Exploiting the principles of Quantum Mechanics and especially those of Quantum Adiabatic Processes, we present the arguments that those noncomputable problems in such a class can be nevertheless computable! About the speaker:
Tien D Kieu came to Australia as a Vietnamese refugee in 1980 and
obained a B Sc (Hons) from the University of Queensland in 1984 and a Ph
D from the University of Edinburgh in 1988. He then held postdoctoral
positions at the Universities of Edinburgh and Oxford, including a
Junior Research Fellowship at Linacre College. In 1991, he took up an
Australian Postdoctoral Fellowship and then a Queen Elizabeth II
Fellowship at the University of Melbourne. At the beginning of 2001,
after 3 years at CSIRO in the position of Senior Research Scientist and
Project Leader, he joined the Swinburne University of Technology as a
Professorial Fellow. He is currently interested in Quantum Computation
and Quantum Measurement.
Topics in Genetic Programming & Reactive Agents This seminar will be in two parts. Part 1: Convergence and Building Blocks in Genetic Programming This talk will present our work investigating the unexpected convergence behaviour of genetic Programming (GP) for classification problems, to be presented at the Australian Artificial Intelligence Conference in December. We attempt to understand why GP classifiers sometimes fail to reach satisfactory levels of accuracy for certain problems regardless of computational effort. Building blocks are speculated to be the fundamental components which GP combines to form accurate solutions, and the building block hypothesis gives a description of how GPs actually work to produce effective solutions. We developed a new method for identifying building blocks within evolving GP runs, and show that some building blocks can lead a GP run to successful completion, whist the use of other building blocks can lead to evolutionary stagnation. Work shows that these building block components have a substantial impact upon the expected success of a GP run. These results give some indication as to the processes at work behind the GP search method and provide empirical support for the building block hypothesis. Part 2: Evolving Fuzzy Rules for Reactive Agents in Dynamic Environments Fuzzy logic controllers have been applied to a wide range of control problems, but are very difficult to build for situations where the environment changes quickly and there is a lot of uncertainty. This work investigates a new method of creating fuzzy controllers, in the form of reactive agents, for such environments. The framework for this investigation is the RoboCup soccer simulation environment, where the agents are in the form of simulated soccer players evolved to exhibit competent dribble-and-score behaviours. The method proposed uses a messy genetic algorithm to evolve a set of behaviour producing fuzzy rules which define the agents. About the speakers:
Tom Loveard is currently a PhD research student at
RMIT in his third and final year of studies.
A New Runtime for the EC Erlang Compiler This seminar will discuss the new runtime being developed for the EC Erlang Compiler, which is intended to be both the core compiler technology for the Magnus massively scaleable computing platform, as well as a means to generate standalone executables of Erlang programs. It has also been used to implement some previously identified safety extensions to improve the safety of Erlang when used for mobile code applications amongst others. These include using capabilities for all critical resources (processes, open files, network connections, references etc), as well as a hierarchy of nodes within an Erlang system. The seminar will start with a brief overview of the Erlang language, which is a functional language designed to support robust, reliable, distributed, near real-time applications, developed by the Ericsson CS labs. Then I will describe the proposed language safety extensions, and provide a brief overview of the Magnus architecture. The latter half of the seminar will discuss my work on the new runtime for EC. This includes the use of detached system pthreads to implement Erlang processes, the use of the dlopen library to implement dynamic module loading, memory allocation for Erlang terms and the consequences for garbage collection, and some open issues on the implementation of capabilities and the use of their rights in the system. About the speaker:
Dr. Lawrie Brown has a position in the School of Computer Science at
the Australian Defence Force Academy in Canberra. He is currently on
sabbatical visiting the Software Engineering Research Centre at RMIT.
The DOGMA Project: Research on a Data-Based Framework for Ontologies and the Semantic Web Formal ontologies can be seen as mathematical objects that form the range of a classical "Tarskian" semantics interpretation mapping of first-order language constructs that could represent situations, functions or procedures related to a given domain. Some design methods and techniques such as view integration that were originally developed for large databases, where the "data models" and their semantics typically are limited to a particular application, could be relevant for this purpose, and we analyze parallelisms and fundamental differences between databases and ontologies. In particular the ORM (Object-Role Modeling) method, or rather its precursor NIAM, with its rigorous distinction and handling of so-called lexical and non-lexical knowledge could prove to be an interesting candidate to help identify and clarify a number of these issues, and I will try to illustrate this with a number of examples. I will argue that in a precise sense, by "externalizing semantics", ontologies allow (database) applications to achieve a level of "semantic independence" much in an analogous fashion that database systems delivered "data independence. The resulting approach leads to the new concept of "double articulation" of ontologies, a term incidentally inspired by computer linguistics. I will discuss how data modeling differs from ontology engineering, but also draw the obvious parallels. I will treat more specifically ongoing research within and around STARLab's DOGMA Project that indicates how such database methods and tools may or must be adapted to be usable in the context of ontologies, and how they may then help to define ontology updates, the role of domain constraints, and future tools that assist in e.g. the alignment of ontologies. I will briefly discuss several case studies arising from a number of past and ongoing EU 5th Framework and other projects that involve ontology technology and methods at their core. About the speaker:
Robert A. Meersman is full professor in the Department of Computer
Science of the Vrije Universiteit Brussel where he leads the
laboratory of Systems Technology and Applications Research (VUB
STARLab) since 1995. From 1986 until 1995 he was full professor at the
KUB (Department BIK) in Tilburg, and prior to that associate professor
in Diepenbeek, Belgium (LUC), part-time at the VUB, and project leader
in the research department of the American computer company Control
Data in Brussels. He teaches database and information system courses,
both introductory and advanced, has been and is promoter of several
PhD theses. VUB STARLab runs several sizeable (multi-person,
multi-year) European and national scientific projects (OntoBasis, FF
Poirot, ImaGen, …) and participates in a number of EU Thematic
Networks (OntoWeb, AgentWeb, CCFORM, InnovaNet). Robert Meersman has
been invited speaker and lecturer at universities and institutes
worldwide, and notably gave the keynote speech at the prestigious VLDB
Conference in Barcelona, 1991.
If you are interested in giving a presentation in this seminar series, or to make suggestions for speakers, please contact James Harland, the seminar co-ordinator. [an error occurred while processing this directive] |