A Proposal for the Evaluation of Adaptive Personalised Information Retrieval

Abstract

Personalisation in Information Retrieval is achieved using a range of contextual information such as information about the user, the task being conducted and the device being used. This information is used to devise the most suitable response for the individual’s need. As such personalised Information Retrieval and response composition approaches become more widely used, traditional evaluation measures become less effective and applicable. This paper proposes that contextual, and specifically personalised, approaches to Information Retrieval could benefit from the experience of the Adaptive Hypermedia community. The paper details the approaches to evaluation commonly used by the IR and AH communities and proposes a means of combining and enhancing these disparate approaches in a unified framework for the evaluation of personalised IR systems.