As computing matures, it is becoming increasingly obvious that a change is necessary for the manner in which web services interact with users. Server-centric models are inconvenient for users. A new paradigm, Smart Interactions, provides a web service architecture which is centered around the user's needs, rather than the simplistic server view currently being used. The system responds to the individual user and is able to adapt to changes to better serve the user. The Smart Internet system helps the user accomplish their tasks efficiently and intuitively. An important aspect of Smart Interactions is that of cognitive support, which provides enhanced information and guidance to the system or user linked to the current task. This thesis examines predictive analytics and its application to cognitive support in Smart Interactions, and presents and evaluates a framework for using predictive analytic support within the Smart Internet model. / Thesis (Master, Computing) -- Queen's University, 2011-09-29 18:11:02.374