Page 45 - Building Digital Libraries

CHAPTER 3

means for systematically identifying high-value resources that users expect,
while avoiding low-value materials that clutter up search results.
Just as it is common in physical libraries to use different selection methods
for different types of resources, it is appropriate to use a combination of
methods to identify materials for inclusion in a digital collection. Selec-
tion is a time-consuming process that requires an understanding of what
resources are available and what practical challenges they present, given the
goals of the repository and the technology it is based on.
For a selection process to be successful, the following questions must
be addressed:

What resources are desired and where are they?
The success of a repository depends on the effective iden-
tification of materials that are relevant to the purpose of the
repository. For this reason, the single most critical aspect of the
selection process is to define as clearly as possible what should
be included and where it can be found. The selection process
must address a number of questions: How will selectors iden-
tify desirable resources? Who is the target audience? Are the
resources in a physical collection waiting to be digitized? Are
they oriented towards a particular topic or purpose? Are they
in a CMS (content management system) or database? Who cur-
rently creates and maintains the resources? Are they part of a
collection now, and if not, are they scattered across multiple
websites? If they are stored in multiple systems, how do selec-
tors know where to find them? What formats are they in? Who
is going to use the collection and for what purpose?
The answers to these questions vary greatly from one digi-
tal repository to the next, so there is no “best” way to identify
resources. If the purpose of the repository is to provide spe-
cialized information on a given topic, the first task is to iden-
tify where the desired information is and who maintains it.
If it turns out that the resources in question are created and/
or maintained by a small number of content providers, it may
prove feasible to ask content providers to manually or auto-
matically help identify useful resources. On the other hand, if
the resources are created and/or maintained by a large number
of individuals with diverse interests, it is unrealistic to expect
consistent assistance in identifying relevant items, so alterna-
tive methods need to be explored.

How will different versions of documents be handled?
Depending on what type of resources a repository includes, ver-
sion control can be a serious issue. Although many resources are
not changed after they are created, digital resources are inher-
ently different from their physical counterparts because they
can be easily modified. If the collection development policy

30