SaS Seminars 2005
Software and Systems Research Seminar Series
Evaluation Dependability Attributes at the Architecture Design Phase
Date: December 6 Tuesday!!!, Place: Alan Turing, Time: 15:15
Dr. Lars Grunske, Boeing Postdoctoral Research Fellow, School of Information Technology and Electrical Engineering, The University of Queensland, Brisbane
Abstract: Over the past decades, software architectures and architecture centric development received increasing attention in the context of complex dependable systems development. These dependable systems need to fulfil requirements regarding dependability attributes such as safety, reliability, availability, maintainability, security and temporal correctness. As these properties be cannot be designed directly into the system, there is also a rising need to predict and evaluate these dependability properties at the architecture design phase. This talk will give overview over the research field, discuss arising problems and challenges and reviews current architecture evaluation approaches.
A Metalogic Formalisation of Legal Argumentation as Game Trees
Date: November 10, Place: Alan Turing, Time: 15:15
Prof. Joergen Fischer Nilsson, Computer Science and Engineering Section, Informatics and Mathematical Modelling, Technical University of Denmark
Abstract: We describe a framework for logical formalization and analysis of argumentation and dispute between two parties. The key idea is to metaphorically consider an argumentation dialogue as a board game like chess. Instead of moving pieces the parties utter speech acts comprising logical propositions in order to defend or attack a key claim. The applied principles are sought demonstrated by unravelling of a simple legal case within statutory law. However, the applied principles are by no means confined to legal reasoning. The dynamics of the argumentation dialogue is achieved conveniently also computationally by adopting a metalogic framework in which propositions are encoded as terms, and by appealing to an inference engine for defeasible logic.
In the presentation we will try to stress the ideas and principles of our framework also for people with little or no background in legal reasoning and/or logic programming, taking the liberty of ignoring some intricacies in the applied logics. To this end we consider the board game tree conception useful.
This is joint work with Andreas Hamfelt and Jenny Eriksson, Uppsala University.
An Industrial Perspective on Technological Challenges in Deep Sub-Micron Design
Date: October 21, Place: Alan Turing Time: Obs!! 13:15
Dr. Björn Fjellborg, Ericsson
Abstract: Highly integrated microelectronics form the platform for signal processing in Ericsson's cellular base stations, with each new product generation presenting ever tighter demands on performance, power, and cost. This presentation gives an industrial perspective on developments in mobile applications, technology, hardware architecture, and design methodology and how they affect Ericsson's implementation strategy. 3G and 4G applications, 65 nm process technology and beyond, multi processor systems on chip, and formal methods for design all enable solutions as well as present challenges. Meeting project budgets and time-to-market requirements and still make the best use of technological opportunities requires a balance between re-use of existing solutions, development of improved architectures, and management of deep sub-micron effects.
Speaker's bio: Dr. Björn Fjellborg received his PhD in Computer Systems from Linköping University in 1992. He has worked in the industry with development of high-level synthesis tools, and since 1995 at Ericsson with digital hardware development for cellular base stations. He holds currently a position as senior specialist in digital ASIC design methodology, and is responsible for the design and verification methodology for baseband ASICs in Ericsson's cellular base stations.
Rational Agents in Logic Programming for the Semantic Web
Date: September 29, Place: Obs Donald Knuth, Time: 15:15
Prof. Luis Moniz Pereira, Director of Centro de Inteligência Artificial (CENTRIA), Departamento de Informática,Universidade Nova de Lisboa, Portugal
We examine rational features employable by agents in the context of the Semantic Web, based on Logic Programming.
In particular, we shall look at the EU project REWERSE -- Reasoning on the Web, -- and address the issues of reactivity and evolution.
Luís Moniz Pereira, Universidade Nova de Lisboa (UNL):
- Full Professor of Computer Science, at Departamento de Informática, UNL, since 1985.
- Director of the Artificial Intelligence Centre (CENTRIA) at UNL
- Elected Fellow of the European Coordinating Committee for Artificial Intelligence in 2001.
- Founding president of the Portuguese AI association (APPIA).
He received the scientific prizes: 'Prémio Estímulo à Excelência' (2005) and 'Prémio da Boa Esperança' (1994), both from the Portuguese government; and 'Prémio de Ciência e Tecnologia' (1984), from Calouste Gulbenkian Foundation.
He is involved in co-ordination of a number of international research projects and international educational programs, funded by the European Comission or bi-lateral, including IST NoE in Computational Logic CoLogNet, REWERSE, Erasmus Mundus supported distributed European MSc in Computational Logic, Asian-Link project and several bi-lateral projects.
Cross-Layer Adaptation for Quality-Aware and Energy-Efficient Next Generation Mobile Multimedia Devices
Date: September 15, Place: Alan Turing, Time: 15:15
Prof. Klara Nahrstedt, Department of Computer Science, University of Illinois at Urbana-Champain
Investigating Software Qualities - the Blekinge Way
Date: May 12, Place: Alan Turing, Time: 15:15
Dr. Mikael Svahnberg, Blekinge Institute of Technology
With the contemporary ubiquitousness of software, where software is seen as strategic assets and where the society becomes more and more dependent on software we are also more susceptible to flaws in the software. Hence, it is of vital importance to develop software with adequate quality. The BESQ project at Blekinge Institute of Technology aims at researching and developing different means for achieving an overall balance of different aspects of quality. In this presentation an introduction to the BESQ project is given, together with a discussion of the challenges involved in developing for and achieving a certain and predictable balance of different quality aspects.
Modular Verification of Software Systems
Date: March 17, Place: Alan Turing, Time: 15:15
Prof. Kathi Fisler, Department of Computer Science, Worcester Polytechnic Institute, Massachusetts USA
Aspect-oriented programming (AOP) has become an increasingly important programming paradigm. AOP is one of a family of means for expressing abstractions that cut across the program's dominant modularity; these techniques enable the construction of software by composing features. Despite their importance, feature-based programming methods lack supporting computer-aided verification techniques, which are necessary for increasing reliability and developing confidence in systems.
This talk presents a technique for verifying feature-based programs (expressed as state machines). By analogy with modular compilation, we support modular verification, both to enable independent development and to reduce the computational cost of verification. While modular verification is an old idea, feature-based programs violate fundamental assumptions underlying standard modular techniques. Using case studies for motivation, I will describe both the subtleties of feature verification and our analysis techniques.
(Joint work with Shriram Krishnamurthi, Brown University)
Kathi Fisler is an Assistant Professor of Computer Science at WPI. Her research centers around formal approaches to understanding and predicting functional behavior of hardware and software systems. She has a particular interest in the role of diagrammatic representations in verification. Her current work explores modular analysis of feature-oriented software systems, the computational advantages of timing diagrams as a logic for temporal specifications, and change-impact analysis for access-control policies. Kathi earned her PhD in 1996 from Indiana University, did a postdoc at Rice University and internships at Bell Labs and Intel before starting at WPI in 2000.
Using reputation-based algorithms to counter the spam problem
Date: February 10, Place: Alan Turing, Time: 15:15
Spam is definitely a major problem, that everybody using email is facing daily. Even though there are quite some spam detection techniques (e.g. rule based), which have significantly reduced the number of spam reaching your mailbox finally, still quite some spam messages remain. Furthermore, there is a permanent struggle between the spammers and the writers of the detection rules so that you are required to keep your spam filter up-to-date permanently.
In this seminar, I will show an alternative distributed technique (MailRank), based on social networks that can be derived from your personal email network. It is shown to perform well in case of attacks (i.e. spammers trying to circumvent the spam detection) since it is based on the PageRank algorithm used by google to rank web pages. In addition to detecting spammers, it can also provide a ranking of non-spam emails as an automatic pre-classification of incoming emails, which is useful for those people receiving many emails per day.
Page responsible: Christoph Kessler
Last updated: 2012-08-17