I would like some feedback on some work Intrallect has done on
implementing a "collection" context set for CQL, and how we can improve
it, and possibly harmonise it with other work in this area.

The aim of Intrallect's "collection" context set is to allow queries to
be limited to sub-collections within a repository.  I would like to make
this into a recognised CQL context set, or adopt an existing context set
which would allow us to do the same thing.

There are some details of what we have implemented so far at
At the moment there is one index, "", used to constrain a
search by collection name.   I am planning to add a
"collection.identifier" index field.

We also want to include information on the collections available in a
repository (i.e. at least the name and ID of the collections).  We
thought we would do this in the SRU explain file, but we have been
struggling to work out what the best extension would be to add this
information to the ZeeRex schema.  Possible schemes include ISAD, EAD
and RSLP, e.g., 
  ISO 2146 or the latest version of MODS.

Ray Denenberg has since suggested using Scan to expose the collection 
information, which is something I hadn't considered, but seems to make a 
lot sense for a CQL index.

Any thoughts would be very welcome.


Martin Morrey, Product Director, Intrallect,
[log in to unmask], Tel: +44 870 234 3933, Fax: +44 1506 505 117