background image
pages of data per month to an audience far
beyond traditional academia. Humanists -
especially those who participate in scholarly
debates that span decades or more - must
think carefully about how they will
respond to this vast new and expanding
audience.We need to ponder both the way
in which we write and the questions that
we pursue. Maintaining the status quo and
dismissing this new audience is itself a
strong, if problematic, response.
The library is a laboratory where
reading is a primary exercise:
To some extent, this is a superset of the
customisation problem. A great deal of DL
research addresses the cataloguing problem.
A digital library is a structured space that
manages a large number of objects.The
user searches through the DL to find
objects of interest, but, once these have
been found, many systems simply hand
control over to the object and the user
calls up a PDF viewer etc. Humanists often
study texts, images and spaces in extremely
close detail.Thus, the numbered citation
schemes of computer science publications -
which direct readers to a document as a
whole - reflect a much less general attitude
to textual reference: humanists are trained
to cite precise pages and, when dealing
with canonical documents, often cite indi-
vidual lines or words. In this environment,
the granularity is much finer and users
need support with words and phrases as
well as with documents as a whole.The
implications are, however, profound for the
scale and design of humanities DLs: when
each word becomes a complex multidi-
mensional object, density of data increases
by several orders of magnitude. Cultural
heritage materials raise challenges that go
beyond those described in the literature
about citation harvesting and linking from
recent scientific publication.
Digital objects and their compo-
nents must be freely reusable:
Simple access to information is not suffi-
cient.We need complex documents that
include and provide distinct visualisations
of components from many sources, e.g.
details from high-resolution images, clips
DigiCULT
.
Info
12
tions have undertaken the crucial task of
digitising some critical corpora and have
produced intellectual gated communities.
These electronic resources, tightly con-
trolled and often priced in such a way as
to guarantee a limited audience, restrict
fundamental source materials to the same
academic elites that had access to scarce
print resources. A socio-economic infra-
structure has thus begun to arise that
imposes on the digital world limitations of
print.We need economic models that do
not replicate practices that isolate cultural
heritage from the community as a whole.
Governmental approaches are, however,
also problematic, since governments may
feel an obligation, explicit or not, to con-
trol their national image and impose
restrictions on information.
Cultural heritage digital libraries
must serve the needs of diverse
audiences:
Access to information is necessary but not
sufficient. Customisation is a rapidly gro-
wing field of inquiry.The system should
adapt to the needs of its users, providing
them with the information that they need
to interpret new documents or topics,
reducing, insofar as possible, the friction of
their movement through a digital library.
There are limits to this - as Euclid repor-
tedly rebuked the first Ptolemy with the
statement that there is `no Royal Road to
geometry',
1
some concepts are simply dif-
ficult. Nevertheless, a humanities digital
library has a social obligation to support
the development of complex skills, by a
wide audience, over a long period of time.
The documents within cultural heri-
tage digital libraries must serve the
needs of diverse audiences:
Humanists cannot simply rely upon
elaborate technologies to enhance their
contributions to society as a whole.The
Internet already reaches a huge audience
and could within a very near future satu-
rate the households of the advanced
countries.The Perseus Digital Library
Website has, for example, emerged as a
major distribution channel within classics
and now disseminates up to 9,000,000
of time-based media, tabular or graphic
visualisations of data sets, quotations from
larger works, and links from each inclu-
sion to the source.
Standards/best practice must be
descriptive rather than prescriptive:
New publication series can impose guide-
lines on the form and structure of docu-
ments.The variations of historical sources
can provide crucial information.We thus
need to preserve, rather than eliminate,
vagaries of spelling in early modern texts
since these variations can provide impor-
tant data about the compositional history
of a given text (e.g. compositors often
provide the actual spelling and uses of `do'
vs `doe,' which, for example, can help
determine who is responsible for what
section of Shakespeare's First Folio).The
need for `prescriptive' rather than `descrip-
tive' encoding demands a consequently far
more complex encoding scheme and soft-
ware infrastructure.This requirement
generates a need in turn for specialised
viewers, which can, for example, filter and
display very precise differences between
editions.While the underlying ideas are
similar to the well-known problem of ver-
sioning source code, a cultural heritage
versioning system requires substantially
more precision of reference and semantics:
editors within the New Variorum
Shakespeare series, for example, formally
distinguish between `substantive' and
`semi-substantive' changes to the text. A
versioning system must be able to manage
a wide variety of such classes.
O
ver the coming year we will be
continuing to shape our understan-
ding and thinking on these issues and to
develop the recommendations to ensure
that the needs of the humanities are re-
flected more strategically in digital library
activities in the US and abroad. But we
have already begun the long road of thin-
king on these issues.
Gregory Crane,
gcrane@tantalos.perseus.tufts.edu
1
Reported by Proclus in his description of Euclid: see
http://ww.perseus.tufts.edu/cgi-bin/ptext?doc=Perseus:
text:1999.01.0086&query=head%3D%232&word=Euclid.)