This requires a disciplined approach applied in a systematic manner to build progressive levels and layers of knowledge.
Attempting to build a single ubiquitous repository of knowledge (federated or otherwise) will not work. To create a systemized modeling capability around knowledge requires the introduction of domain models. In this case, we are applying domain modeling to the problem space to create finite spaces for decisions, content, context, inputs (direct & indirect), and outputs (direct & indirect).
The direct & indirect inputs and outputs are purposely stated to separate inputs that are “first class citizens” in the problem domain versus those which are tangential (those which may influence and/or augment the decisions within the problem domain). Similarly, the indirect outputs are influences (triggers or data) which may impact other problem domains through a series of cascading events.
A direct input is something that must be in place in order to properly guide the decision process. A direct output is something is triggered or produced by the problem domain based on one or more branches of decision logic.
Examples: Problem Domain - Purchasing
Examples: Problem Domain - Contract Management
Examples: Problem Domain - Asset Management
Examples: Problem Domain - HR (Benefits Management)
Examples: Problem Domain - Vendor Management
Examples: Problem Domain - Capital Equipment Maintenance