The CITE architecture for identifying and retrieving objects

Overview

Any architecture for a long-term networked library of interoperating digital publications needs to address at least three levels of design, each with a very different life span:

TierLongevityExamples of technologies
End-user application Very short lived (a few years?) Web page using HTML + JavaScript for retrieving asynchronous content (AJAX)
Service Longer life (multiple generations of technology) XML exchanged over HTTP
Reference or citation Permanent Simple string with defined syntax and semantics (e.g., URN)

The range of material we need to support is summarized in the acronym CITE:

We can break down the architecture as a matrix of designs for each type of material at each tier. For the lower two tiers of network services and citations, it looks like this:

The CITE matrix

TierCollectionsIndexesTexts(an example of) Extended Objects
Service Collections service Reference index service Canonical text services CHS image extensions
Reference or citation DNIDs n/a CTS URN Image citation with Regions of Interest