For what it's worth, we've been discussing this kind of "intelligent linking" feature in detail lately among the PKP group, and it seems like a natural fit with our work as part of the Synergies
project. Your reference to the ACM portal and click-through is exactly the kind of feature we'd like to support, and at the moment, it looks like the OpenURL system (something we've considered embedding in the past) is the front-runner for our approach.
The full OpenURL specification
is somewhat lengthy, but there are a few overview papers
; basically, it can contain any set of metadata available (although 1.0 spec is based on the Dublin Core profile) -- so, if DOIs are available, they can be used, but they are certainly not required. We definitely want to implement a system that doesn't require subscribing to any particular ID framework, while supporting those who are using them.
Linkback systems are somewhat limited in that they aren't really designed to support the kind of rich metadata that scholarly citations require; but the approach is sound and might make a good conceptual starting point. We also have the Open Archives Harvester
, which might be a good candidate to add OpenURL resolution functionality to. The main challenge is maintaining authority in a decentralized approach, and by building OpenURL linking into the PKP suite, we'd be providing a self-assembling framework that also supports an existing bibliographic standard.
It's still in early discussion stages right now, but one way or another we'll certainly be building this kind of functionality into upcoming releases.