Friday, 14. January 2011 14.01.11 12:35 Age: 7 Jahre

Using internet resources efficiently
Junior professor Dr. Patrick Briest develops algorithms for combinatorial pricing

By: Katharina Bätz

The internet offers junior professor Patrick Briest ample room for research. (Photo: Thomas Thissen)

Paderborn. Patrick Briest has been at the Institute for Computer Science at the University of Paderborn since September 2009. Originally from Bottrop, Germany, Briest is a member of the research group on algorithms and complexity and focuses especially on the interaction between autonomous software systems. Before coming to Paderborn, Mr. Briest spent time at universities in Dortmund, Liverpool and New York.

A key element of Patrick Briest’s work concerns the application of algorithms and game theory principles on large, fractioned systems, of which the internet is an example. The internet is a platform that allows for many different autonomous systems to interact, each system primarily following its own interests. To achieve a functioning overall system, the protocols governing the interaction of the individual actors need to provide incentives to collaborate. At the same time, because of the immense size of the system, the algorithms used need to be as efficient as possible, in order to work within the limited computing and storage capacity of the system.

"Game theory was developed in economics, and we are now looking into the question of how game theory and efficient algorithmic solutions can be combined and applied to fundamental problems", explains Patrick Briest. One application, on which he is working together with others at the Institute for Computer Science and colleagues in the USA, is the search for efficient algorithms for combinatorial pricing – which plays a role in the placement of advertisements on search engines.

"In essence it is about the age-old problem of finding result-maximizing prices for a given set of products; a problem that is older than the internet", says the junior professor. To make this possible, data concerning the preferences of potential customers need to be gathered and analyzed. Based on this data set, algorithmic optimizations can be developed. To be able to guarantee efficiency, approximate or random approaches are used.

"With the help of these methods, online merchants for example, can find out how much money they can still demand for a specific product", explains Patrick Briest. "Because it is the case that online pricing is much more variable than in the offline world, since it cannot be checked", he adds.

Another interest of the computer scientist is the widespread use of alternative sales methods on the internet.

Pricing strategies enabling price discrimination as, e.g., different types of product bundling, are employed widely as a means to boost revenue. Concrete examples range from hotel bookings on platforms like hotwire.com to the variety of complex cell phone plans offered by different service providers.

The development of algorithms is one of the oldest branches of computer science. The research at the Paderborn Institute for Computer Science focuses on the question of whether current technical possibilities (such as mobile communication networks or high performance computing) offer new avenues for the development of efficient algorithms.

Contact:
Jun.-Prof. Dr. Patrick Briest
Research group: Algorithms and Complexity
05251 60-6457
patrick.briest[at]upb.de