User modelling and adaptivity in visual information retrieval systems
José Manuel Torres · 2002
This paper proposes suitable characteristics of an ideal Visual Information Retrieval (VIR) system. The central role of user modelling in such a system is discussed. Approaches to the automated analysis of images in image retrieval systems are considered. A preliminary sketch of the model is then provided. The model is based on Bayesian user modelling techniques and dynamic Bayesian Networks. A brief worked example of the model is provided, showing how the model might acquire knowledge about the user and exploit automatically derived image features to help satisfy a user’s specific image retrieval requirements. 1