Page 34 - Building Digital Libraries

Choosing a Repository Architecture

• A repository can support only a limited number of formats,
and emulating environments is sometimes not technically
or legally feasible. In the physical world, libraries collect,
maintain, and preserve resources only when they have the
capacity to do so. This principle is still important with digi-
tal resources.

How Will the Repository Be Managed?

People are a critical component of every repository—however capable the
platform, individuals make the decisions and do the work that determine
what can be added to the repository, how it is managed, and how it can be
used. For any repository, the following questions must be answered:

• Who provides oversight for the repository as a whole?
• Who decides what content will be added?
• Who oversees the incorporation of resources into the
repository?
• Who creates content?
• Who is responsible for individual resources and collections?
• Who supplies technical expertise to maintain the repository
and advise on the management of specific resource types?
• Who addresses curatorial issues that emerge over time?
• Who manages compliance to ensure that access and reten-
tion are handled properly?

Other High-Level Platform Decisions

A number of other high-level decisions need to be made before choosing
a repository architecture. High-level architecture decisions as to whether
the repository should be optimized for specific types of resources, based on
open source or proprietary software, hosted locally or remotely, and based
on virtual or real machines, must be based on users’ and institutional needs
rather than on broad generalizations.

Special-Purpose or General Platform?
Different products are optimized for different purposes for the simple reason
that they were initially created to attain different goals. For example, con-
sider the three early platforms adopted by libraries: DSpace, CONTENTdm,
and Bepress. All three theoretically provide access to all types of resources,
but each is much better suited for some purposes than others because they
were designed to perform different tasks and are based on fundamentally
different assumptions.
DSpace was initially conceived as a way to disseminate research through-
out user communities, and it is still best suited to that task. According to this
19