Topic:	Efficient, self-contained handling of identity in peer-to-peer systems
Date:	November 3, 2005, 13:00 to 14:30
Place:	Room SR -108, AVG (Allgemeines Verfügungsgebäude, 50.41), Adenauerring 20a, 76131 Karlsruhe
Abstract:	Identification is an essential building block for many services in distributed information systems. The quality and purpose of identification may differ, but the basic underlying problem is always to bind a set of attributes to an identifier in a unique and deterministic way. Name/directory services such as DNS, X.500, or UDDI are a well-established concept to address this problem in distributed information systems. However, none of these services addresses the specific requirements of peer-to-peer systems with respect to dynamism, decentralization and maintenance. We propose the implementation of directories using a structured peer-to-peer overlay network and apply this approach to support self-contained maintenance of routing tables with dynamic IP addresses in structured P2P systems. Thus we can keep routing tables intact without affecting the organization of the overlay networks, making it logically independent of the underlying network infrastructure. Even though the directory is self-referential, since it uses its own service to maintain itself, we show that it is robust due to a self-healing capability. For security we apply a combination of PGP-like public key distribution and a quorum-based query scheme. We describe the algorithm as implemented in the P-Grid P2P lookup system (http://www.p-grid.org) and give a detailed analysis and simulation results demonstrating the efficiency and robustness of our approach.
Speaker:	Dr. Manfred Hauswirth is a senior researcher at the Distributed Information Systems Laboratory of the Swiss Federal Institute of Technology in Lausanne (EPFL) since January 2002. He holds an M.S. (1994) and a Ph.D. (1999) in computer science from the Technical University of Vienna. Prior to his work at EPFL he was an assistant professor at the Distributed Systems Group at the TU Vienna. His main research interests are on large-scale distributed systems, peer-to-peer systems, sensor networks, self-organizing systems, Web services, e-commerce systems, and publish/subscribe systems. He has published over 40 papers in these domains, he has co-authored a book on distributed software architectures and several book chapters on P2P data management and semantics. He has served in over 80 program committees of international scientific conferences and recently was local chair of WDAS2004, program co-chair of SME05, program co-chair of STD3S, and program co-chair of MCISME. He is a member of IEEE and ACM. His hobbies are music, reading, travelling, skiing, and since very recently - golf.

Topic:	How to aggregate preferences without revealing them
Date:	November 17, 2005, 12:30 to 14:00
Place:	Room 253, Building 11.40
Abstract:	Whenever a group of autonomous entities, such as humans or computational agents, strive for a global decision, they need to aggregate their possibly conflicting preferences. Well-known examples of traditional preference aggregation among humans are auctions and voting. Besides achieving a socially "desirable" outcome, agents are interested in maintaining the privacy of their individual preferences. We investigate whether privacy that does not rely on trusted third parties can be achieved in preference aggregation. More precisely, we study the existence of distributed protocols that allow agents to jointly determine the outcome of an aggregation mechanism simply by exchanging messages and without revealing any unnecessary information. This is called "private emulation" of a mechanism. We provide results for both the computational (or cryptographic) and the unconditional (or information-theoretic) setting. It turns out that, in the unconditional setting, some mechanisms cannot be emulated privately at all and that the construction of efficient protocols for the computational setting is a very challenging task. Common types of sealed-bid auctions will be used as standard examples throughout the talk.
Speaker:	Felix Brandt is a junior research group leader in the Computer Science Department of the University of Munich (LMU). He holds a diploma (1998) and a Ph.D. (2003) in computer science from the Technical University of Munich (TUM). Prior to his work at the LMU he was Post-Doc research fellow in Prof. Sandholm’s Agent-Mediated Electronic Marketplaces Group at Carnegie Mellon University from 2003 to 2004 and in Prof. Shoham’s Multiagent Group at Stanford University from 2004 to 2005. Felix Brandt is principal investigator in the PAMAS research group, i.e. a group devoted to studying distributed protocols that allow autonomous entities to aggregate their conflicting preferences. The lab is hosted by the chair for theoretical computer science at the University of Munich and is funded by DFG (German Research Foundation) within the Emmy Noether Program (Aktionsplan Informatik).

Topic:	Combinatorial Auctions - An Introduction
Date:	December 1, 2005, 14:00 to 15:30
Place:	Room 231 (Building 11.40)
Abstract:	The internet infrastructure allows for new trading mechanisms that enable participants to express complex preferences. Examples are bundle trading in combinatorial auctions or multi-attribute auctions. The talk will provide an overview of combinatorial auction formats, applications and design problems.
Speaker:	Martin Bichler received his MSc in Information Systems from the Technical University of Vienna, and his Ph. D. as well as his Habilitation (2000) from the Vienna University of Economics and Business Administration, where he was Associate Professor at the Department of Information Systems. From 1997 to 1998, he was a research fellow at UC Berkeley. 2001 to 2003 he was research staff member in Deep E-Commerce team of the IBM T. J. Watson Research Center, Yorktown Heights, New York. Martin Bichler has been involved in research and development in the areas of supply chain management, electronic markets, distributed systems, and operations research. He has published papers in journals such as Decision Support Systems, EJOR, GDN, Computer Networks, and the Journal of Distributed and Parallel Databases. He has designed and developed several software packages. He was lead architect and project manager for the first version of MyTravelDream.com. The IBM MAP object framework has been integrated in Digital Union's ezMarket Platform and is now in use with a number of large-scale procurement auction sites and private exchanges. Martin Bichler has been on the program committee of many international conferences and workshops and is on the editorial board of the Electronic Commerce Research Journal, Manufacturing and Service Operations, the Journal of Production and Operations Management, the International Journal on Web Services Research, and the International Journal of Business Process Integration and Management. He is a member of the ACM, GI, WKWI, the IEEE Task Force on Evolutionary Computation in Finance and Economics, co-chair of the IEEE Computer Society Technical Committee on E-Commerce (TCEC) and the Task Force on E-Commerce of the German Informatics Society (GI FB 5.5). He is board member of the Bavarian Elite Master program "Finance and Information Management".

Topic:	Introduction to Software Agents and E-commerce Agent Systems
Date:	December 15, 2005, 14:00 to 15:30
Place:	Room 216 (01.80, Wachterhaus)
Abstract:	Since 1994 we are supposed to believe that software agents will become the next revolution in computer science. This change is to occur not only in the ways we construct software but it is also to have a much broader impact on the field of human-computer interaction. Unfortunately, the revolution prophesized in 1994 does not seem to materialize. When we turn the computer on in the morning, we do not contact “our agent” to receive a personalized newscast, our day-plan and – on the basis of that plan, the weather forecast and knowledge of our dressing-preferences - an advice what to wear. Similarly, when creating software for an e-shop we do not utilize pre-existing agent-modules. To the contrary, it is rather difficult to point to a successful large-scale implementation of an agent system completed using one of the multitude of existing and constantly created (and fading away) agent environments. The aim of the lecture is three-fold. First, an introduction to software agents is presented with a discussion of major points raised “for” and “against” software agent systems. Second, it is shown, that the implementation of large scale agent systems is possible, as state-of-the-art agent platforms (e.g. JADE) easily scale up to over a thousand agents and 300.000 messages. Finally, a positive research program is presented and illustrated by a model agent-based e-commerce system using negotiating agents with dynamically loadable modules.
Speaker:	Marcin Paprzycki works as an Associate Professor at the SWPS University in Warsaw, Poland. He has received his M.S. Degree in 1986 from Adam Mickiewicz University in Poznań, Poland and his Ph.D. in 1990 from Southern Methodist University in Dallas, Texas. His initial research interests were in high performance computing and parallel computing, and over time they evolved toward distributed systems and Internet-based computing; in particular, agent systems. He has published more than 200 research papers and was invited to Program Committees of over 200 international conferences. He is on editorial boards of 8 journals and book series. Maria Ganzha works as an Assistant Professor at the Elbląg University of Humanities and Economics in Elbląg, Poland. She obtained MS and her Ph.D. in Applied Mathematics from Moscow State University in 1987 and 1991 respectively. After devoting her time to her two daughters she has recently returned to active research and works in the areas of software engineering and agent systems. She is on editorial boards of 2 journals and was invited to program committees of almost 20 conferences.
Slides:	Download(need Login first)

Topic:	Fairnet: A Reputation System for P2P Data Structures
Date:	Januar 26, 2006, 11:00 to 12:30
Place:	Room 348, Informatik-Hauptgebäude (50.34),
Abstract:	Peer-to-Peer (P2P) Data Structures (a.k.a. P2P Overlays or Structured P2P Networks) distribute the data set and the workload among many peers in an open, large-scale environment. Most of the approaches for P2P data structures assume that all peers readily follow the protocol and carry out its share of the workload. But the economic dominant strategy is to refuse to work off queries issued by other peers, i.e., to behave uncooperative. The talk will describe a reputation-based protocol that renders uncooperative behavior unattractive. The protocol exchanges feedback between adjacent peers that acknowledges performed or refused work. In addition, the talk will address the challenges that arise from spoof feedback and present the results of a detailed algebraic and experimental evaluation.
Speaker:	Erik Buchmann