FCS Agenda
This is to collect ToDos? and Tasks regarding Federated Content Search.
MannheimMeeting? on 11/12/2012
Update after Meeting in Nijmegen 2011-08-24
The basic search-result format is (more) fixed. Changes:
<ccs:Metadata>
element is dropped in favour of<ccs:DataView type="metadata" >
- New attribut:
@schema
on element<ccs:DataView>
- Attributes @pid and @ref on any element
- New
<ccs:DataView type="title" >
encouraged to use wherever possible
New questions:
- Really distinguish between
@pid
AND@ref
? If yes, allow both at the same time on one element? Which one has precedence than? - Relation between
DataView
-attributes@type
and@schema
. - Relation between
@type
-attribute and mime-type? Proposal for@type
being a superset of mime-type, i.e. mime-type values can be used in @type, next to some other basic values specified by the specification (kwic, title, metadata, content
).
2. Specification of DataViews
In the next step we need a definition of the actual format of the individual DataView Types. This has to be harmonized with the work on Viewers. A separate page Viewable elaborates on this relationship.
Interested teams should take over individual types and lead working out a proposal. Tentative assignements:
- text, annotated text, written corpus, book
- Thomas@UniTueb; Matej@ICLTT
- list, matrix
- Thomas@UniTueb
- geolocation
- Marc@Meertens
Implement Search on Indexes
After the basic connectivity is established and simple queries can be submitted, the work shall be started on allowing to query specific indices. This involves solving the announcement of indices via explain/scan and correctly parsing the non-trivial query:
index relation searchTerm
Search context
Elaborate on ways of selecting a search context (Repositories/Collections/Resources to content-search in) either extensionally by providing a list of identifiers or intensionally by MD-query. See more on SearchContext and CDMDC.
Use cases
We need concrete scenarios to be able to check if what we do is any useful.
Further test Pazpar2
Assigned: Matej, ICLTT
There is a test-instance of PazPar2 running, that now can already connect to FCS-services. Currently main issue is Pazpar2 not being able to pass through XML-data inside the customizable fields.
Establish Repository Registry
For now we collect the information about existing services under RepositoryRegistry but this needs to be made a separate service.
How do we describe the Repositories: endpoint-link in Collection
-MDRecord vs. CMD-Profile Repository
and/or explain
-record.
(see: RepositoryRegistry#IdeasonimplementationoftheRegistry and CDMDC#LinkingMDRecordsandContentServices)