Graphical models represent conditional independence relationships between variables, including, for example, those between the various symptoms and causes of a disease. An important topic in the area of machine learning is learning these types of models from data. In some applications, it is crucial to include information that is not contained in the data, i.e. prior information. The aim of this research is to design an efficient algorithm that utilises prior knowledge in a manner which allows users to express what they know about the problem domain. This involves creating a system where the input is composed of prior knowledge, together with data, connected to a Bayesian learning algorithm. Our main aim is to facilitate the design of an algorithm that uses prior knowledge ahead of time, in order to both speed up the process of learning and ensure that the learning is more accurate.