A workflow pattern could be defined as the methods by which a user typically utilizes a particular system. This paper presents a framework by which data mining techniques could be used to extract patterns from an individual's work flow data to exploit a type of architecture known as a Knowledge Advantage Machine (KAM). KAM is a type of semantic desktop and semantic web application that would assist people in constructing their own personal knowledge networks, as well as sharing that information in an efficient manner with colleagues using the same system. A KAM would be capable of automatically discovering new knowledge that is relevant to the user's personal ontology. Through experimentation, it is empirically demonstrated that a user's file usage patterns can be utilized by a software to automatically and seamlessly learn what is "important" as defined by the user. Further research is necessary to apply this principle to a more realizable KAM so that decisions can be fueled by work patterns as well as semantic or contextual information.