Design principles for e-journal archiving
The target open source application is intended to be deployed in libraries as a
lightweight “Smart Client” that can be installed on most desktops. A “Smart Client” is
an application that optimises resources locally and through web services connectivity,
and can operate in both an offline and online manner. Development frameworks and
techniques supporting Smart Clients have matured considerably since the first
concepts of Smart Client emerged in 1997 (Yoshikawa et al., 1997).
The project is designed to capitalise on Smart Client design approaches to achieve a
blend of local and networked operation in environments where technical skills are not
necessarily common. There is clear applicability of the design concepts in this
proof-of-concept model to the deployment of library services applications in developing
countries, where similar resource and connectivity constraints apply. The following
considerations framed the application architecture:
. To provide a single install process, and involve minimal or no configuration.
. The application must install and deploy all components within its own
framework (e.g. it must not rely on other external components such as Apache).
. The user interface must be unambiguous, with the least amount of pages
possible and with a common structure across all pages.
. The system must be able to run on an average specification PC.
. The user interface must display on multiple monitor sizes.
Key functional elements of the local element of the application are:
. The ability to draw application and metadata updates from web services
providers.
. The system must be a configurable archive that builds a store of the journals that
exist in the current electronic subscriptions. The system must also provide a
means for viewing the archived content.
The client maintains an independent archive, stored locally on the computer that it is
currently running on (Figure 1). The client acts as an autonomous agent, crawling the
journal sites based on the user configured subscriptions, gathering the full text journal
for local storage. It implements a generic spidering pattern that has been developed to
support multiple journal provider web site structures through the use of provider
specific spidering rules, which are not built into the client, but rather accessed from the
metadata server through the use of web services. Not only does the metadata server
provide the spidering rules, it also provides an index of the journals offered by each
supported provider.