Attendees: Farrukh, Kathryn, Nikola, Carl, Reagan
Agenda:
1. Approval of Minutes (from June
22):https://lists.oasis-open.org/archives/regrep/201206/msg00007.html
Approved
2. Minute taker (volunteer needed - last minute takers: (Farrukh,
Farrukh, Farrukh, Farrukh, Oliver, Farrukh, Richard, Farrukh,
Kathryn, Oliver, Farrukh, Kathryn, Nikola, Farrukh, Oliver, David,
Farrukh, Kathryn, Farrukh, Carl, Farrukh, Nikola)
Farrukh
3. Review of proposed CQL
Profile:https://lists.oasis-open.org/archives/regrep/201208/msg00004.html,
https://wiki.oasis-open.org/regrep/documents/plan/regrep4/CQLProfile
- Farrukh: Reviewed background and context for Image and CQL
Profile proposals as described in
https://lists.oasis-open.org/archives/regrep/201208/msg00004.html
- Farrukh: Reviewed CQL Profile proposal:
https://wiki.oasis-open.org/regrep/documents/plan/regrep4/CQLProfile
- Uses existing AdhocQuery query with CQL as the query
language and a CQL query string as the queryExpression
parameter. No new query or query protocol changes are needed!
- Farrukh: CQL came from Z3950 and being standardized by OASIS
SearchWS TC
- Nikola: We had tried collaboration with SearchWS TC before,
Is there a need to do it again?
- Farrukh: Last time we proposed to have them use our
protocol. This time we want to use their query language. There
is no strong reason for a formal collaboration other than to
have normative reference to their CQL [SearchRetrievePt5]
specification in our CQL Profile
- Farrukh: In future it would be possible for any RegRep
extension profile spec to define one or more CQL contexts,
each defining context-specific search indexes that can be used
in CQL queries as well as context -specific relations that may
be used to match indexed data. Image Profile spec is the first
such spec that uses CQL to enable ad hoc querying of
profile-specific data in an interoperable way
- Reagan: Need a way for clients to discover what contexts are
supported by a server
- Farrukh: The SearchWS protocol defines how. We need to
specify that RegRep server support similar capability somehow.
Will add to Issues section
- Reagan: There are 3 levels of abstraction: the query language
(e.g. SQL, CQL, SPARQL), the public protocol (e.g. ebRS),
private (internal) protocol between RegRep server and remote
data source. Example of latter is Integrating with an OpenDAP
server that returns NetCDF format data. Need some sort of
protocol mediation between public protocol (ebRS) and internal
protocol (OpenDAP)
- Farrukh: There is nothing in RegRep spec to define this
today. Currently it is in implementation space. If there is a
proposal for a spec to address this we should discus it in a
focused way in a future meeting
4. Review of proposed Image
Profile:https://lists.oasis-open.org/archives/regrep/201208/msg00004.html,
https://wiki.oasis-open.org/regrep/documents/plan/regrep4/ImageProfile
- Farrukh: Reviewed Image Profile proposal:
https://wiki.oasis-open.org/regrep/documents/plan/regrep4/ImageProfile
- EXIF and IPTC specs are quite complex and large
- Currently only defining a small subset of metadata
properties in EXIF and IPTC as indexes for the Image CQL
Context
- The Image CQL Context is likely to evolve as we gain
experience with more indexes and relations
- Cataloger map EXIF and IPTC metadata that have mapping to
Image CQL Context to slots that follow a naming convention
that matches the index name in slot.name. The slot ValueType
matches the data type for the EXIF or IPTC property
- Nikola: Should we allow for people to add additional indexes
for ImageProfile?
- Farrukh: that would be non-interoperable. Need to decide on
all supported indexes as part of spec. Implementations could
do more if they wish but thats outside the spec.
5. Additional items?
Next meeting: October 12, 2012 9AM PT / 12PM ET
--
Regards,
Farrukh Najmi
Web: http://www.wellfleetsoftware.com
|