Theory for Searching in a Thesau- rus

In this article we present a framework based on Situation Theory which models the Query by Navigation process in a Hypertext environment. Descriptors in the Hyperindex are transformed to so-called infons. Documents in the Hyperbase are represented as situations, which can support infons. Similarly, search paths are modeled as a number of followed information links between the infons expressing the information need of a searcher. Based on the user's search path two preference relations are proposed. Firstly, a preference relation for documents, which is expressing a kind of relevance ranking. This ranking is based on a support relation and on an amount of history. Secondly, a preference relation for descriptors is suggested. This ranking can be used as a support for the user. This article shows that formalizing the decisions of a search process in a neutral setting can gain insight into the behavior of the user, consequences for the relevance ranking, and the properties of the search support.