Preservation of Intelligibility of Digital Objects - Advanced Digital Preservation - page 162

Fig. 8.13 Modelling DC profiles without making any assumptions

their dependencies exist in the DC profile. For example Fig. 8.13 showssomeofthe
dependencies of a file in FITS format. The users of profile u know how to run the
module FITS S/W since all the modules it requires exist in T(u) . However they
cannot decompile it, even if they know the module FITS S/W , since we cannot
make the assumption that a user u will also understand JAD and its dependencies.
This scenario requires changing the definitions of some intelligibility-related
because the set of modules understandable by a user u is its profile T(u), and not
Nr ∗ (T(u)). Intelligibility checking and gap computation are now defined as:

Deciding Intelligibility: True if Nr + (t)

T(u)

Intelligibility Gap: Nr + (t)

\ T(u)

Another consequence is that we cannot reduce the size of DC profiles by keeping
only the maximal elements since no assumptions about the knowledge are permitted.
In the case where dependencies are disjunctive, we do not make any assumptions
about the knowledge that is assumed to be known, since the properties of the various
tasks and their dependencies are expressed explicitly. In this case the performability
of a task is modelled using two intentional predicates. The first is used for denoting
the task i.e.

IsEditable(X) :- Editable(X,Y).

and the second for denoting which are the dependencies of this task i.e.

Editable(X,Y) :- TextFile(X), TextEditor(Y).
DC profiles contain the modules that are available to the users (i.e.
TextEditor (' NotePad ').). To examine if a task can be performed with a mod-
ule we rely on specific module types as they have been recorded in the dependencies
of the task, i.e. in order to read a TextFile X, Y must be a TextEditor.
However users may know how to perform such a task without necessarily classi-
fying their modules to certain module types or they can perform it in a different way
than the one that is recorded at the type layer. Such dependencies can be captured by
enriching DC profiles with extensional predicates (again with arity greater than 2)
that can express the knowledge of a user to perform a task in a particular way, and