Understanding search

This thesis provides a framework for information retrieval based on a set of models which together illustrate how users of search engines come to express their needs in a particular way. With such insights, we may be able to improve systems’ capabilities of understanding users’ requests and through that eventually the ability to satisfy their needs. Developing the framework necessitates discussion of context, relevance, need development, and the cybernetics of search, all of which are controversial topics. Transaction log data from two enterprise search engines are analysed using a specially developed method which classifies queries according to what aspect of the need they refer to