While there have been significant advances in capturing data from the entities
across complex real-world systems, their associations and relationships are largely
unknown. Associations between the entities may reveal interesting system-level
properties that may not be apparent otherwise. Often these associations are hypothesized
by superimposing knowledge across distinct reductionist representations of
these entities obtained from disparate sources. Such representations, while useful,
may provide only an incomplete picture of the associations. This can be attributed
to their dependence on prior knowledge and failure of the principle of superposition
in general. Such representations may also be unhelpful in discovering novel undocumented
associations. A more rigorous approach would be to identify associations
from data measured simultaneously across the entities of interest from a given system.
These data sets or digital signatures are quantized in time and amplitude and
in turn may (dynamic) or may not (static) contain explicit temporal information.
Symmetric measures such as correlation have been helpful in modeling direct associations
as undirected graphs. However, it is well appreciated that the association
between a given pair of entities may be indirect and often mediated through others.