COSI - Summer School on Decision Aiding
23-24 Mai 2009
Université Badji Mokhtar, Annaba, Algeria
Sponsored by Microsoft Research through Microsoft Inspire Program
Tutorial Summaries
I) Mashups, SaaS, and Cloud Computing: Evolutions and
Revolutions in the Integration Landscape
Professor Boualem Benatallah University of New South Wales, Sydney, Australia
II) Working with Preferences: Why, when and how
Doctor Souhila Kaci, CRIL-CNRS 8188, IUT de Lens, Université d'Artois
III) Business Intelligence: Essentials and Economic Challenges
Professor Aris M. Ouksel, University of Illinois at Chicago, USA
IV) Propositional Satisfiability :
recent advances and challenges
Professor Lakhdar Saïs, CRIL – CNRS UMR 8188, Université d’Artois
V) Parallel optimization and financial mathematics
Professor Pierre Spiteri - IRIT / ENSEEIHT - UMR CNRS, France
VI) Bioinformatique, Intelligence Artificielle et réseaux
de régulation géniques
Professor Laurent Trilling, University Joseph Fourier, Grenoble I, France
I) Mashups, SaaS, and Cloud Computing: Evolutions and Revolutions in the Integration Landscape
Boualem Benatallah, University of New South Wales, Sydney, Australia
This tutorial will be offered as a seminar at IEEE ICDE'2009, Shangai (http://i.cs.hku.hk/icde2009/).
Integration is a key technique in software engineering, which aims to bring together disparate
components and systems to form new, value-adding applications. In this context, web mashups,
software/platform/infrastructure as a service, and cloud computing are novel, innovative paradigms
and forms of integration that are fascinating a rapidly growing number of researchers and practitioners.
Yet, the exact meaning and scope of those terms, the technological challenges underlying these paradigms,
as well as the research and business opportunities they bring are still vague and sometimes hard to grasp.
This seminar aims at clarifying these paradigms, at discussing the relationships that exist among them,
and at outlining the fundamental challenges and potentials they bring.
II) Working with Preferences: Why, when and how
Doctor Souhila Kaci, CRIL-CNRS 8188, IUT de Lens, Université d'Artois
It is commonly known that preferences are very useful in many real-life problems. They guide humain decision making from early childhood (e.g., "which ice cream flavor do you prefer") up to complex professional and organizational decisions (e.g., "which investment funds to choose?"). Preferences are inherently a multi-disciplinary topic, of interest to economists, computer scientists, operations researchers, mathematicians, logicians, philosophers, and many more. Preferences are a relatively new topic in Artificial Intelligence and are becoming of greater interest in many areas such as non-monotonic reasoning, multi-agent systems, constraint satisfaction problems, decision making, etc. In this course, we first overview and compare different preference representation languages. Then we revisit some preference-based frameworks (multiple criteria decision, database queries, constraint satisfaction problems) and show how they can profitably benefit from the recent advances on preference representation in Artificial Intelligence.
III) Business Intelligence: Essentials and Economic Challenges
Professor Aris M. Ouksel, University of Illinois at Chicago, USA
Organizations, both private and public, feel increasing pressures to capture, understand, and harness their internal and external data to support decision-making in order to improve their operations, to support quick contingent responses to rapidly changing conditions, and to be innovative in their operations. Legislation and regulation (such as the Sarbanes-Oxley Act of 2002 in the USA) is requiring their leaders to document their organizational processes and sign off on the legitimacy of their information they rely on and report to stakeholders. Compression of cycle times due to globalization of the economy requires organizations to be agile and make frequent and quick strategic, tactical and operational decisions, some of which are extremely complex. Managers need the right information at the right time and in the right place. This is the essence of modern approaches to Business Intelligence (BI). We will first discuss the major characteristics of BI today and tomorrow, and business performance management. Then, using our current work on the impact of price discrimination and market segmentation on competition and consumer purchase behavior, we will address another key question, which is how can enterprises in competitive environment maximize their BI investments.
IV) Propositional Satisfiability : recent advances and challenges
Professor Lakhdar Saïs, CRIL – CNRS UMR 8188, Université d’Artois
Propositional satisfiability (SAT) is the problem of deciding whether a boolean formula in conjunctive normal form (CNF) is satisfiable. SAT is one of the most studied NP-Complete problems because of its theoretical and practical importance. Encouraged by the impressive progress in practical solving of SAT, various applications ranging from formal verification to planning are encoded and solved using SAT. Research in SAT started in the context of automated theorem proving, where SAT was identified as a simple instance of formally proving theorems. Theorem proving is regarded as a subfield of artificial intelligence. Many problems in AI can benefit from practical and theortical advances in SAT solving. In the last decade many advances in SAT were driven by the electronic design automaton (EDA) community with their huge interest in efficiently solving large SAT instances. The growing need for more efficient and scalable verification techniques have fueled research in verification methods based on SAT solvers. In this course, after an overview of recent advances in satisfiability are presented. Then we motivate and explore the recent development in parallel Satisfiability. Applications and challenges of SAT technology are finally presented.
V) Parallel optimization and financial mathematics
Pierre Spiteri - IRIT / ENSEEIHT - UMR CNRS, France
For a mathematical problem, derived from financial mathematics, we set up a link between the formulation of the specific problem and an optimization problem defined on an adapted space with respect to the economic specification of the studied application. The term of the classical Euler conditions leads to the solution of boundary values problems; more specifically, we have to solve evolutive variational equalities or inequalties according to the economic context. The previous problem derived from the Euler conditions are solved by parallel synchronous or, more generally, asynchronous iterative algorithms for which the convergence is analyzed by various ways. Finally, results of parallel experiments carried out on various multiprocessors architectures and grid network will be presented; particularly the parallel synchronous and asynchronous iterative algorithms are compared.
VI) Bioinformatique, Intelligence Artificielle et réseaux de régulation géniques
Professor Laurent Trilling, University Joseph Fourier, Grenoble I, France
On le dit couramment : l' "informatique sera à la biologie ce que les mathématiques ont été à la physique". Après un tour d'horizon sur les grands problèmes (bio)informatiques posés par la biologie cellulaire moderne, nous aborderons une classe de problèmes dits post-génomiques, c'est-à-dire dépassant le cadre des séquences présentes dans un génome : la modélisation des réseaux de régulation de gènes. Ce sera l'occasion d'étudier précisément un formalisme discret proposé par René Thomas pour modéliser ces réseaux et de montrer comment une vision de type Intelligence Artificielle appliquée à ce type de réseaux conduit à la conception de systèmes informatiques originaux. Nous illustrerons ce propos à l'aide d'un réseau présent dans la bactérie E.coli.