Collection browsing

Owned by Bill Branan
Jun 16, 2017






Whilst search and discovery using facets is a very powerful approach, we still have users who want to be able to browse around the underlying repository structure. Our original repository implementation was structured in sets so that, for instance:

Electronic dissertations and theses

  • Adult Eduvation
    • PhD
  • American Studies
    • PhD
  • Applied Biology
    • PhD
      ...
  • Postgraduate Medical Institute
    • ClinPsyD
    • MD
    • PhD
    • PsyD
  • Psychology
    • ClinPsyD
    • PhD
    • PsyD
  • Scandinavian Studies
    • PhD
      ...
      etc

These implicit sets are related in our Fedora RELS-EXT datastreams by 'isMemberOf' relationships and we have a number of tools that use the RDF to build and display trees for navigation.

Whilst we are not actually 100% sure that we want to implement this, this is what we envisage for our Hydra IR head at present:

Modified home page

  • This is the normal home page but with a link at bottom left that switches to browse view

Home page for browse

  • The facets have vanished and been replaced by a link. Instead we have a panel showing the top level sets in the structural tree of the repository (generated from the repository RDF).

Open level 1

  • clicking on a top level collection (in this case ETDs) opens the next level...

Open level 2

  • clicking on, say, Adult Education opens the next level...

Search return

  • clicking on PhD, which is a set with actual content (rather than more sets) returns a list.

Three levels of nesting like this is as far as we support at present.

This drill down approach is still at the planning stage but, whilst it certainly won't be in Phase 1, we'd maybe like to see it in Phase 2 (summer 2011). During a period of user testing in the spring, we are going to see how many people really want this and consider further what are the use cases. We have an inkling that carefully structured facets presented in a careful order might be enough to enable most users to mimic the approach they used in the 'old' repository. That would leave a very small number of collections that could potentially be dealt with using explicit (rather than implicit) collection objects.