Knowledge Discovery from Data in the Context of Prior Beliefs

Knowledge Discovery from Data (KDD) allows business, science and government to derive significant value from their data assets. However, "discoveries" that are already known to the user will not be useful and can mask those that have value. This project will create user-centric KDD technologies that extend the state-of-the-art to take account of a user's existing beliefs. We will invent approaches for modelling a user's beliefs, using data to find violations of those models, assessing the relative potential value of those violations and communicating those violations to the user. This will enable the discovery of much more useful knowledge than found by current methods, together with more effective support for exploiting that knowledge.