INFORMATION RETRIEVAL BASED ON AXIOMATIC DECISION THEORY
S. K. M. Wong, Peter Bollmann, Yiyu Y. Yao · International Journal of General Systems · 1991
Information retrieval systems are designed with the objective to identify useful information items in a very large collection of documents in response to a user request. In such systems, one has to deal with two contentious issues, the representation of documents and the construction of queries. In this paper, based on the axiomatic decision theory we present a framework to study these complex issues. Two quantitative representation systems for different preference structures are introduced. One is based on the expected utility model and the other is developed from the notion of belief functions. An inductive method is also provided for constructing queries from user preferences.