The following list gives an overview over my research projects. The projects are characterized by a short abstract and are linked to a corresponding web-site which provides further information.
Pricing of Distributed Information Services (PRISE):
In a world of distributed information systems, as it is represented in the Internet, the efficient allocation of resources within the system is a critical coordination task. Unlike traditional approaches using scheduling algorithms, PRISE is focussing on the economic interpretation of resource allocation problems. The assignment of priorities to tasks by using the consumers willingness-to-pay has been a first step into this direction in this context. This project was supported by the German Research Society until its completion in 2003.
Dynamic Bundle Pricing (PREMIUM):
The objective of this research project is the development of a software tools that enable practitioners to build software applications for the automated allocation of resources and bundles of goods especially in distributed computer systems such as Grid systems. The research in this project employs combinatorial auctions and yield management techniques in connection with methods of artificial intelligence, such as reinforcement learning and neural networks. The project is funded by the German Federal Ministry of Education and Research since 2003.
Financial Business Grid (FINGRID):
Increasing competition in the banking sector leads to a high pressure for automation in the IT-related business processes of the German financial industry. New legal regulations such as Basel II and a customer demand for highly customized products enhance this pressure. The FINGRID project strives to identify suitable services and processes in the financial service industry and to develop Grid-based systems that enable the financial service providers to reorganize their IT infrastructure and to realize applications that have been unthinkable without Grid technology. Economically organized Grid systems are a new and promising research topic in this context. To guarantee relevance, the research will be performed jointly with industrial partners, such as Deutsche Bank, Dresdner Bank, Finanz_IT, IBM, DataSynapse and PA Consulting. Grounding on the technology developed by the D-Grid Initiative, applications for service provisioning will be developed by integrated pricing and accounting structures into the Grid. The project has started in May 2007.
FINancial Agent-based Computational Economics (FINACE): Bounded rationality, heterogeneous agents and herding effects are the most important ingredients in constructing evolutionary adaptive systems to simulate experimental security markets. The intention of the FINACE project is part of the debate about whether expectations affect asset prices and may lead to excess volatility. The relation between capital market anomalies and bounded rational behaving investors is the main focus of this study. For this purpose a new experimental market "Frankfurt Artificial Stock Market" (FASM) will be implemented as an open and modular system to better understand the interactions between the heterogeneous acting agents in a trader network and the resulting dynamics of the pricing process.
Collaborative Optimization of Last Mile Networks for Courier, Express, and Parcel Delivery Services (KolOptNet):
Rising energy costs and fierce competition urge carriers to increase transportation network efficiency. KolOptNet focuses on creation of sustainable, energy efficient logistics by designing algorithms and implementing methods for optimized network and tour planning within parcel and express delivery networks. KolOptNet integrates dynamic tour planning heuristics and combinatorial auctions to ensure the long-term participation of the service providers by using incentive compatible distribution mechanisms for the cost savings. The project makes use of optimization methods that use hybrid metaheuristics including ant algorithms as well as local neighborhood and tabu seach.
Economics of Networks in Information Systems (EONIS):
Recent research on network topology in real-world systems has shown that there are three different basic classes of networks: random, small-world, and scale-free. These topological properties are claimed to have direct impact on the economics of systems based on such a network. The project tries to identify the network structures in several application areas of the information system domain and to quantify and explain their impact on their economic behavior. The software `SVN Network Analysis Tool' (SVNNAT) which is part of the project aims to reveal the strength and the properties of collaboration ties between the developers in order to analyze their productivity and their quality of work in a given constellation of a software development network.
Real World Games (REWOG):
The founders of game theory (Borel, Morgenstern, von Neumann, Kuhn, Nash and Shapely) were involved in the study of "real world games", like Poker. In the meantime the interest in the study of classical games has dramatically reduced due to the widespread purely mathematical analysis of representative strategic situations in the main stream of game theory. The study of classical games like Poker, however, offers the opportunity to study the interconnection of psychological elements, such as "bluffing", the purely mathematical analysis of the game itself, and the variation of playing strategies by means of artificial intelligence methods. It is the combination of the psychological aspects of a game and the mathematical analysis of the strategies contained, that makes REWOG interesting. In a first step we developed the "Texas Hold'em Fixed Limit Heads-up" bot PLICAS based on the Computer Poker Competition platform. PLICAS uses the tool RangeCalulator to find the optimal bluffing frequency according to the opponet's playing style.