CLARIN-FCS specification: issues and notes for next revision

This page is a non-exhaustive list of issues that should be addresses in or some notes for the next CLARIN-FCS specification.

Form and Content

• (brought up by Olha) Hit highlights in Generic Hits Data View: it's not specified how hits should be highlighted if an query contains boolean expression, e.g. kind AND taal:
  • Should endpoints highlight the whole hit? Or should they highlight only the matching words, thus creating discontinues hit markers?
  • Should the behavior be controlled by the client, e.g. by sending a auxiliary parameter like fcs-spans?
  • Jörg: We should do as other search engines do: Use different highlighting colours for the the variables (in the example above: kind, taal) and highlight them (this gives discontinues hit markers)
• CQL clarification for FCS: quoted strings shall be handles literally; unquoted strings are invalid. Note relation to MWE below.

Implementation

• (brought up by Olha) XML-Schema design: always add Schema types, e.g. for EndpointDescription? element. This allows easier use of toolks like JAX-B.
• Agreement on how to treat MWEs. After a poll with the technical contacts of the endpoints, there is an agreement that unquoted MWEs should be treated like being implicitly quoted. Thus the aggregator will be changed in one of the comming releases to not send unquoted MWEs to the endpoints as a first measure. This will be enforced in client libs for FCS version 2.

Discussion

None yet.