Notes from London F2F about BP1

(scribed by @jtandy as a complement to the official minutes)

BP1: Include spatial metadata in dataset metadata

• @byron is concerned that this is overly weighted toward DCAT
• … the relationship between “full-fat” metadata (e.g. an ISO19115 compliant geospatial metadata record used within an SDI) and more “friendly” metadata like DCAT [needs to be better understood]
• … main issue is the monolithic presentation of a typical full-fat metadata record
• … DCAT provides an “exoskeleton” for wrapping metadata
• … are there examples of this in the wild?
• @clemens: no comment on DCAT, just that ISO19115, DCAT and schema.org are _all_ approaches that should be considered
• … DCAT is only really seen in Government e-portals - there’s no indication that DCAT has wider traction
• @ed: we need to look at a range of approaches
• @byron: interested in preserving the native metadata, but exposing that through things like DCAT
• … the practice “behind” the suggestion of DCAT (!)
• … e.g. “when you’ve got existing metadata from your SDI, re-use this in a Web-friendly way”
• … the full-fat metadata has been created for a reason, just like you would use schema.org if you want to make stuff discoverable on the Web by search engines
• @phil notes an upcoming DCAT workshop
  • … [which will consider] content negotiation by profile - allowing users to access metadata according to their intended goal / purpose; e.g. to negotiate retrieval of ISO19115/DCAT/schema.org based on what the client application needs
  • … there is an RFC in progress; @larsG is involved
• @andrea: DCAT should see wider use - it is easily extensible (e.g. GeoDCAT-AP profile) to meet needs of a wide range of communities … wider than geo
• … it allows metadata from multiple domains to be published in an accessible way, allowing evaluation of the dataset
• @ed: we make data more accessible; but don’t throw away the [dataset] metadata … find a way to make that _metadata_ more accessible too
• @linda: metadata is for different purposes - for geospatial, the most important is [spatial] extent (for discovery; e.g. “where” is the data), then resolution etc. (for evaluation; e.g. does this data meet my needs?)
• … so for discovery, use schema.org
• … use DCAT to judge if the data is useful … “evaluation”
• @jeremy / @payam: don’t distinguish between dynamic data streams and published datasets; e.g. a sensor data stream updated in real time vs. a curated cadastral dataset published annually - they both need spatial ‘dataset’ metadata for discovery & evaluation
  • … need to say this in the BP doc
• @phil: the key for DCAT is the concept of _distribution_ vs _dataset_
• @clemens agrees; this would answer his concern about BP1
• [also see discussion around BP17 and the conclusion to add section on _data provenance_ into BP1 and include crowd-sourced dataset as an example, potentially just a single attributed #uksnow tweet]