Traditionally, the knowledge acquisition process used to populate a knowledge-base involves two
people (or groups of people): an expert in the domain and a knowledge engineer. The human expert
provides information to the knowledge engineer, who then encodes it as data in the knowledge-base.
However, due to the dynamic nature of daylight and the large number of design variables that can
affect performance, it would be difficult or impossible for a human to accurately provide the information
required to populate the knowledge-base. For our system, a set of simulations has been used which
allow us to obtain highly specific data. The following sections describe the methodology used to
obtain this data.