background image
DigiCULT
.
Info
42
F
urther development should be trans-
ferred to third parties. Content provid-
ers can suggest materials and ideas, but the
expertise to develop and maintain appli-
cations should come from the educational
sector and educational publishers.
GOOD PRACTICE AND LESSONS
LEARNED
T
he first phase of the project will last
until the end of 2004. At the moment
it is too early to draw any conclusions from
our experiences, especially regarding prac-
tical use and demand. However, there are
some aspects of the process that can already
be commented on:
Scanning costs
Scanning costs should go down further
to make large-scale digitisation of cultur-
al-historical resources feasible. The large
diversity of materials and the special treat-
ment demanded by the unique materials
cause staff costs to remain high. The costs
of encoding moving images have decreased
considerably. The increased use of broadband
requires a higher image quality, which will
increase the production and storage costs.
Standards
Memory of The Netherlands uses a sim-
ple metadata scheme, which accommodates
all descriptive and semantic metadata that
is available in each organisation. The search
form, developed for a wide audience, offers
the search options `who', `what', `where',
`when' and `search all fields'. In many cases,
only the last option will yield acceptable
results, since the structure and the content
of the underlying metadata are heterogene-
ous and inconsistent.
A
ll metadata is stored in a Dublin
Core-like format. Conversion to DC
(extended) is planned for the near future.
The same holds for the version of the SRU
protocol that is used. By upgrading these,
we will meet international requirements for
interoperability.
Metadata
The creation of descriptive metadata is
costly, even when fairly minimal. The qual-
ity of metadata is often poor, ambiguous
or completely lacking. The development of
tools to generate metadata (semi-)automati-
cally seems feasible for text-based materi-
als, but comparable aids for image materials
are still in their infancy. The heterogene-
ous nature of data formats and methods for
semantic-based access that Memory uses
allow for little precision. To standardise this,
and to enrich the metadata by intellectual
effort only, is not feasible. In due course, IT
tools will have to provide a solution.
Process Control
For the final result of digitisation projects
it is worthwhile to invest in monitor-
ing the quality of the process. Digitisation
of heritage materials does not differ much
from other project activities, but co-opera-
tion between content experts and tech-
nical developers is essential. It pays off to
make an extensive analysis of the materials
(objects and metadata) during the prepara-
tory phase and to document precisely every
stage of the digitisation process. In the end,
this will contribute to both efficiency and
quality.
Integration of heterogeneous materials
Making heterogeneous materials cross-
searchable yields considerable advantages
to users. This integration should be real-
ised mainly on the level of the metadata.
Storage, subject description, and presenta-
tion of the various types of materials each
require specific methods, which can be
accommodated only in a modular way. In
the Memory, text and images have been
dealt with in similar fashion. This limits the
options for searching and presentation of
texts. For this reason, a separate trajecto-
ry has been chosen for newspapers, where
the integration will be realised only in the
end-user interface. Text materials in the
Memory (a growing component) will be
revised next year to allow for a better pres-
entation.
THE FUTURE
I
n the future, the Memory's functionality,
i.e. the possibilities to search and retrieve
the objects, will gradually be improved and
extended. Furthermore, a scalable infra-
structure has been realised that can be re-
used by future digitisation projects. The
extensive knowledge and experience that
will have been gained and documented
will result in a professional organisation,
clear agreements, and improved efficiency.
The government is willing to support, and
continue to support, the Memory of The
Netherlands as an integrated part of the
digitisation framework in The Netherlands.
Both end-users and other parties will ben-
efit from this.
Weest dapper, wordt Stormer (Be brave, be a Stormer)
Used with permission of NIOD (Nederlands Instituut voor Oorlogs
Documentatie)