My research is primarily divided into two major foci: developing tools and techniques for practical strategic reasoning (empirical game theory) and applying the developed methodology to market games.
In cases where exact analytical models of a game cannot be constructed, empirical game-theoretic models often provide a viable alternative. I address three basic problems of interest within empirical game theory. First, how should strategies be evaluated in an empirical game model? Second, how should modelers optimally select strategy profiles to simulate? Finally, what game model best predicts payoffs resulting from agent play?
Internet advertising provides a substantial source of revenue for online publishers, amounting to billions of dollars annually. Sponsored search is a popular form of targeted advertising, in which query-specific advertisements are placed alongside organic search-engine results. The placement (position) of an ad for a given query, along with the cost (to the advertiser) per click (CPC), is determined through an auction process. Under cost-per-click pricing, both the publisher and advertiser bear some of the risk associated with uncertain user behavior. The use of automated auctions addresses the combinatorial problem of quoting an appropriate price (CPC) for each display slot for each distinct query. Advertisers bid for the family of keywords of interest, and competition among them determines the going CPC for each of the available slots on a query-by-query basis.
The management of supply chains poses a complex strategic problem faced by many of today’s businesses. In response to the growing complexity, many firms engaged in various information technology services such as EDI systems and ERP services. The business processes involved in and surrounding these services are becoming increasingly automated. Analyzing the efficacy of the mixed initiative and autonomous systems regulating supply chains is becoming an increasingly important research question. The Trading Agent Competition Supply Chain Management scenario addresses many interesting aspects of such a challenging problem in a repeatable simulation environment.
As part of the STIET Summer Research Opportunity Program, I mentored students participating in the second annual TAC/CAT competition tournament. Undergraduates Kemal Eren, Flavio Kuperman, and Bongseog Choi extended the MANX (Michigan Agent for Negotiated eXchange) from the 2007 competition. The much improved agent placed second in the 2008 competition.