We can see from both of these examples that one single source of material lends itself
to multiple use and interpretation, and one system of use and interpretation can rely on
multiple sources and ownerships (hegemonies). Every new observer or questioner of a
system will compound the problem, and there is no guarantee of stability. When we view
a knowledge system as a communicative process (Walsham, 2005) (embedded in both the
understanding and expectations of the practitioners and the mechanisms being created
to meet those expectations), we can model the system at the teleological level,
aggregating the needs that can be anticipated, and modelling the entirety as a series of
questions that are going to be asked of such a system when complete. In other words,
when we model a system for knowledge retrieval we have to model the flow of questions
and answers that exist within that system. By modelling the questions and answers a
system needs to provide, we can plan allocation of question-answering (QA) resources –
we can delegate to different infrastructures the questions that are best suited to it,
including outsourcing complex queries or work out what is best suited to a reference
librarian or a consultative expert.