ubiquity, heterogeneity and multifunctionality of metadata, this test identified 137
different metadata assertions (i.e., different instances of types of metadata), and 16
types of assertions. Two types cut across all stages of the lifecycle, one cut across two
stages, and the other fifteen were evidenced only in one stage. (Gilliland et al., 2008,
p. 31).
In 1999, the Australian Recordkeeping Metadata Schema (RKMS) concisely
identified the objects that are the primary focus of archivists and other
recordkeepers as a set of entities such as records, recordkeeping, agents, business
process and mandates as well as the various relationships that might exist between
each at any point in timespace (McKemmish et al.). The developers of the schema
assert that these are the objects that need to be described at relevant points in their
lives, from the moments of systems design and records creation onwards. The task of
that description is to ensure that these objects can be understood semantically and
epistemologically, and trusted bureaucratically, juridically and socially. Additionally
they must help the user to assess the object with regard to its authoritativeness,
authenticity and reliability; they must place it within its broader documentary
context or archival bond (e.g., within a given fond); identify it (and here it would be
interesting to look at the degree of consonance with ideas developed in the LIS field
of cognitive authority as well as of 'relevance' and 'aboutness' (Wilson, 1983;
1968)); and support potentially infinite interpretations and understandings of what
the objects being described reveal about dynamic constructions of identity, memory,
and truth.
All of these abilities are, of course, in turn contingent upon the degree, nature and
ultimately the trustworthiness of the associated metadata, so much so that the
management, elimination and preservation of digital metadata through some form
of trust regime has become a growing professional concern (Gilliland, Rouche,
Evans and Lindberg, 2005). If trusting a record means the existence of trustworthy
metadata, how then does one manage constantly accruing metadata in such a way
that it maintains its trustworthiness? This may prove to be one of the most pressing
questions facing these fields in the face of escalating amounts of metadata in the
digital world and also growing political distrust in records and data. One suggested
approach is to develop a metadata management scheme whereby redundant
metadata would be eliminated and notarized summaries would be prepared for large
aggregations of metadata, which could then also be discarded. Another possible
approach would be more ecological-relying upon organizations to invest more in
sustaining the trustworthiness of those records and other objects that they wish or
need to trust and preserve. However both approaches raise important ethical issues
related to equity and the resource capacity of different institutions and communities
to manage relevant necessary metadata over the long-term that meets professionally
and legally acceptable practices for preserving its authenticity.
One additional observation should be made at this point. As Wallace predicted,
metadata has proven to be essential in recordkeeping, and to have more and more
potential as new digital capabilities emerge, but it is not everything. Recent
developments in digital forensics rely not on metadata but instead on very physical
properties of the medium on which records are inscribed. Nevertheless, interpreting
the trace inscriptions detected through digital forensics also must draw upon
existing trusted contextual metadata.
Finally, although the concept of metadata in the archival science and other
recordkeeping fields has been defined primarily in the context of electronic records
- and theoretical developments regarding born-digital materials have largely been
focused on records created by bureaucratic activities and systems - it is important to
bear in mind, as stated at the outset of this paper, that metadata is inherent to all
information and cultural objects. Electronic records research and development have
helped us to understand the roles and manifestations of metadata, but everything
we have learned can be applied to records and other archival materials regardless of
their genesis or form. For example, Yeo makes the point that organic accumulations
and artificial collections may not be mutually exclusive aggregations, even though
theory and practice tends to draw quite a rigid distinction between the two (Yeo,
2010; 2012a; 2012b). Artificial collections, as well as aggregations of personal
papers are also the products of human activity and also have inherent metadata that
speak to their genesis, nature, transmission, aggregation and so forth. Moreover, as
we move forward and use value-added linking metadata such as RDF, as well as
automatically exploiting and compiling inherent metadata elements in archival,
digital humanities and big data initiatives, we will be moving into a new universe of
meta-fonds and meta-collections and meta-metadata.
Characteristics of Archival and Recordkeeping Metadata
Notwithstanding Eno's admonition that metadata is too complicated to explain,
at this point we can observe that at its most straightforward, archival and
recordkeeping metadata can be thought of as encompassing anything and
everything that relates to an object of interest within these fields (i.e., any entity or
relationship) that is not the object itself. This metadata accumulates and envelopes
the object as it moves through time. Metadata also operates in various one-to-one
and one-to-many relationships. For example, metadata can relate to an individual
object, or to an aggregation of objects. Structural, epistemological and ontological
relationships exist between different types and instances of metadata, in addition to
the relationships between the kinds of entities identified in RKMS. For example,
there are multiple organic and cumulative relationships between the provenancial,
descriptive, preservation, interpretative, affective, performative, and 'making' and
're-making' metadata layers that accumulate around and between the original
record 'objects' as well as their versions and copies and the fonds within which they
are filed. Originals, copies and parts of a record can also participate in new archival
formations, with various familial-type relationships (parent-child, sibling, spouse,
cousin, generational) that potentially have certain resonances with the notion of
bibliographic families first articulated by Wilson (1986). Digital capabilities only
expand the ways and moments in and at which this can occur.
We might further consider dividing metadata into two types. The first we might call
'trace metadata', that is metadata that is indigenous to the records and the system
and agents that created and handled them. Trace metadata is often created
simultaneously with the actions that it reflects, and accumulates across actions, for
which it serves as evidence. That evidence might be called upon in legal and audit
processes, for institutional decision-making and self-knowledge, and for historical
analyses, among other uses.
archives in liquid times
218
anne j. gilliland 'the wink that's worth a thousand words': a contemplation on
the nature of metadata and metadata practices in the archival world
219