confirming the agenda
SimonCox: issue on UoM for byteSize still outstanding
Action: SimonCox to contact Alejandra about the UoM
<trackbot> Sorry, but no Tracker is associated with this channel.
SimonCox: wasGeneratedBy was added to the list of properties applicable to Dataset
… the proposed Project ontology isn't mentioned directly because specification is out of scope
… Any comments?
… at 5.8 there is a trimmed down version of the example I used earlier
… I've also drawn attention to the other starting-point PROV properties and we need to decide if they are going to be included in the normative spec
… the full chapter remains to be written
… This proposed execution of the resolution from last week's meeting should be reviewed over this week and discussed in full next week
proposed: accept minutes
SimonCox: DaveBrowning is concerned that we don't have owners for the issues and this is required to push ahead on the second draft
roba: the default position is that the Chair has responsibility for open issues
SimonCox: we are close to closing #1. Closing #2 will allow us to close #3. We will end up with c. 5 that still need effort, and they are large issues
… My view is that the list is aspirational, and we will make a second draft anyway - the strategy we took for the first was we were showing direction. We can do the same for draft 2 so that people have a target to comment on. I'm relaxed that we will get the majority of the items, and any left out will be included in the next sprint
DaveBrowning: I'm OK with that approach
SimonCox: we now have an issue based on this comment
… included in issue #259
… We should deal with it as part of that issue
SimonCox: This comes from the Australian Data Commons
… there are some issues relating to how we are proposing to deal with services
roba: this one comes down to whether or not the client would realise that the endpoint is used in different contexts
… if the data is a catalogue then it is a data discovery service
<SimonCox> Need to clarify that a single URL/endPoint might be both a download + discovery + transformation service
roba: but if you characterise the catalogue as a dataset then it will have the same endpoint. The interpretation is a matter of perspective
SimonCox: the 3rd point is about the general relationships between datasets - something that we need to still fully work through
roba: the point about catalogues is that they can be viewed as data sources, but there isn't a distinction of their behaviour as a data source vs as a catalogue
… it comes down to the profile/s supported
SimonCox: in the overview figure in the doc, one of the things we realised from Genoa is that a catalogue is a special case of a dataset - in Fig1 it is a subclass
… but I've not fully thought through the inheritance of all the properties. We should supress some - esp those inherited from Resource
<SimonCox> some housekeeping required to acknowledge inheritance of properties from Resource
SimonCox: going back to the original question, we need to complete the documentation about the Resource and DataService where the narrative is not yet completed
<SimonCox> zero to many resources? https://github.com/w3c/dxwg/issues/81
SimonCox: Finally, on the cardinality , it is related to #81
… we still need to complete this
… in the 'bag of files' work we included hasPart initially, but removed it for clarity and to avoid clash with the prov:derivedFrom
roba: there is more than one way to do this, but we need to provide guidance on the 'best' way
SimonCox: in the pull request I've made 3 changes
roba: examples will strengthen the guidance - we need to help people with recursive patterns
<DaveBrowning> +1 to providing strong steer in guidance..
SimonCox: all the issues covered in this last comment in the public list are now in the issue tracker
<SimonCox> - finer-grained semantics
roba: One of the challenges with finer grained semantics is the bottomless pit aspect, and someone will feel that it doesn't cover exactly what they want
… a lot of it comes down to compliance with the requirements of the particular project.
… use of dct:conformsTo is a useful extension point, but anything further is a point of pain
SimonCox: are you roba suggesting that this is in the profile?
roba: we need to be clearer about the role of application profiles to give a recommendation about using this route to declare interoperability settings. We will find it difficult to make any other recommendation
DaveBrowning: i agree with roba . in the other issues we have needs for describing finer-grained semantics, but I'm worried that discussion on subsetting might go the same way (bottomless pit)
… I agree with the basic direction, but we need to watch that if we say only a little about finer grained semantics, we might not be able to say much about other things
… it is not yet obvious that we have done enough to decouple subsetting, versioning and finer-grained semantics
… I am happy with the approach , but we need to pay attention to the consequentials
<Zakim> SimonCox, you wanted to ask if we can provide any guidance about specializing dcat:theme?
roba: where we have first-class concerns they will capture info on semantics. we need to point to the key semantics captured in some of these issues (subsetting, etc) and we need to make a list of recommendations but make it clear where people are "off piste"
SimonCox: I put together a wiki page early on - data cube, observation level, etc. there is a diagram (application profile) which is cute, and makes sense, but it needs to follow the general approach for profiles that we are developing
… Do we have/Should we include something like this diagram?
roba: because we are discussing the profiles of the dataset rather than dcat, so the dataset profile should cover this, not the profile of dcat
SimonCox: the properties added to Dataset are the ones people use to search, but in dcat there is only Theme or Keyword
roba: describing your dataset and identifying those elements that are brought up into the catalogue is something that it would be worth developing coherent guidance on
SimonCox: makx has always said 'more documentation', and this diagram is an illustration
roba: i think it is a good idea to write up as an example of how to use DCAT, and also to point out that it is also a way of describing the dataset itself in the profile doc, and we need to ensure that the profile doc uses the same illustration
roba: there should be consistency between the discovery part and the data description part w.r.t. canonical vocabulary etc.
roba: from an alternative perspective, the RDF cube vocab operates at a different level of abstraction - it is a reflection mechanism, talking about the properties in the dataset
SimonCox: the sosa properties might never be mentioned in the data
roba: in the cube the dataset description gives the dimensions and controlled vocabularies used, and might be the only one in the W3C canon that allows description of the dataset structure
+1 to the requirement for consistency and the focus on end user re: vocabulary etc
Succeeded: s/full example/full chapter/
Succeeded: s/Not me.//