The estimation of information value will be a key issue for upcoming intelligent information systems, that not only provide information on request but deal with time-dependently varying information needs. Due to its subjective nature, information value cannot be computed and each individual has a different perception of what exactly is relevant to him. However, we propose a concept of information value and a computational model for its estimation. Our model is based upon the observation that information concerning the future can easily be divided into relevant and irrelevant information as long as the future is deterministic and well known. We use a network representation of uncertain quantities to model the uncertainty of the future. Information value is introduced as the likelihood of reaching a situation in the future where the information is relevant. The approach is demonstrated for the modelling of individual travel planning.
