A framework for the evaluation of adaptive IR systems through implicit recommendation

Abstract

Personalised Information Retrieval (PIR) has gained considerable attention in recent literature. In PIR different stages of the retrieval process are adapted to the user, such as adapting the user's query or the results. Personalised recommender frameworks are endowed with intelligent mechanisms to search for products, goods and services that users are interested in. The objective of such tools is to evaluate and filter the huge amount of information available within a specific scope to assist users in their information access processes. This paper presents a web-based adaptive framework for evaluating personalised information retrieval systems. The framework uses implicit recommendation to guide users in deciding which evaluation techniques, metrics and criteria to use. A task-based experiment was conducted to test the functionality and performance of the framework. A Review of evaluation techniques for personalised IR systems was conducted and the results of the analysed survey are presented. © 2011 Springer-Verlag.