Niche formation and efficient learning of consumer preferences in a dynamic information economy.
Brooks, Christopher H.
2002
Abstract
In many multiagent systems, agents are required to solve a generalization of the <italic>connection problem</italic>, locating other agents that they want to interact with. This can require an agent to learn about the other agents in a population. When this learning is costly, a tradeoff results between the quality of the connection and the time needed to discover this connection. In this thesis, we examine the problem of computational agents attempting to solve the connection problem within the context of an information economy. This leads us to an examination of niche formation, where an agent that is selling information goods must decide to which subset (or <italic>niche</italic>) of the consumer population it wishes to sell. We study this problem at several levels. Initially, we consider a monopolist interacting with a consumer population and ask how it can best learn what prices to charge. Different models of the consumer population approximate the consumer demand to different degrees of precision. When learning is costly, an agent must trade off the profit captured by different models of a population against the time needed to learn these models. In many cases, an agent must also consider other agents that are simultaneously learning. We examine this problem within the context of two producer agents that are simultaneously learning price schedules. When producers selling identical goods can compete only on price, a price war emerges. When producers can differentiate themselves through the use of different schedules and the consumer population is sufficiently heterogeneous, the decision as to whether to engage in a price war can be formulated as a Hawk-Dove game. When producers must learn and learning is costly, niches become even more attractive compared to a price war. A key to producers being able to successfully target separate niches is their ability to differentiate themselves. One way to do this is through the selection of categories of goods to offer. When producers compete in category space, the incentive to target separate niches, rather than to act as generalists and try to appeal to the entire population, depends upon both the way the consumer population evolves and the cost consumers incur for determining a good's value. We then turn to the problem of niche formation in the large, when many producers and consumers are learning simultaneously. We refer to this as <italic> congregating</italic>. We present a model of congregations and characterize the difficulty of congregating, both in the affinity group domain and in an information economy. When agents must pay a search cost in order to allocate goods, congregating both improves net profit and allows a multiagent system to scale to large numbers of agents.Subjects
Consumer Preferences Dynamic Economy Efficient Electronic Commerce Information Learning Multiagent Niche Formation
Types
Thesis
Metadata
Show full item recordCollections
Remediation of Harmful Language
The University of Michigan Library aims to describe library materials in a way that respects the people and communities who create, use, and are represented in our collections. Report harmful or offensive language in catalog records, finding aids, or elsewhere in our collections anonymously through our metadata feedback form. More information at Remediation of Harmful Language.
Accessibility
If you are unable to use this file in its current format, please select the Contact Us link and we can modify it to make it more accessible to you.