Dataset Exchange Working Group Teleconference

alejandra: not yet completed but there is a spreadsheet
… with a different classification of requirements
… also looking at other versioning requirements

<alejandra> spreadsheet classifying the requirements per entity: https://‌docs.google.com/‌spreadsheets/‌d/‌1csCcOEWF4Rh-ghIjxaqKQ4qryfjiQlIpvV6_aFDJn5k/‌edit#gid=113335154

alejandra: can copy the relevant mail into the wiki
… some requirements require fine grained metadata
… those are probably under-specified
… need to follow on regarding application profiles
… not very much in the UC regarding versioning, still going
… through these and will write to the list

Caroline_: https://‌www.w3.org/‌2017/‌dxwg/‌track/‌actions/‌45

dsr: still open

Caroline_: https://‌www.w3.org/‌2017/‌dxwg/‌track/‌actions/‌49 assigned to Annette
… Annete is not present, we'll discuss next week

<Caroline_> https://‌docs.google.com/‌spreadsheets/‌d/‌16JmtNCz_aCWtTCSntriDWLvyPY2x-Y9dZFhAHFl55r0/‌edit#gid=1131839466

UCR

Caroline_: Looking at headings

roba: sent a list of suggestions for headings
… for ordering the spreadsheet
… a few related to alignment to DCAT and other standards
… including equivalences to schema.org
… a number are related to the issue of citation and managing identification
… in the case of federation (multiple copies)
… some relate to fine-grained machine-readable descriptions
… some to profiles
… and others to the fitnesse for usefulness in a specific case
… and some more not quite worked out yet (action constraints etc.)
… should be reasonably coherent

kcoyle: thinks that we should look at how those interact with alejandra's proposal

roba: that's fine, the substantial issue is to get the requirements right

kcoyle: yes, the two spreadsheets should inform each other and figure out where there are gaps

alejandra: we can explore both and they should be complementary.
… Rob's is more topical, alejandra's is more about common aspects among entities
… it's good to have the flexibility to have different classifications

Caroline_: agrees with alejandra's proposal to explore both

<Caroline_> 6.13 Describe fine grained semantics and structure of datasets

roba: OGC provides data exchange schemas, in W3C there is SHACL etc
… constraints on use of vocabularies etc
… which is all about fine grained semantics
… Also W3C QB

<alejandra> https://‌w3c.github.io/‌dxwg/‌ucr/#RID11

roba: so there is some evidence that fine grained semantics is a concern
… in order to understand the data
… and have a formal description of it.
… But there is no standard way to attach that semantics to the data and tell people which vocabularies we use to attach the descriptions

alejandra: one of the examples whe has referred to before since the title of the requirement is to generic
… should be more about dimensions and vocabularies.
… Suggests to go back to the Use Cases and look at the terminology used there
… and the requirements should be refined to better mirror what's in the use case
… perhaps split into several requirements

kcoyle: agrees with alejandra , should be more specific
… perhaps rather a function for the profile
… or a link between DCAT and the profile
… This information might be too community-specific and thus out of scope for DCAT

roba: agrees with kcoyle re community specificity
… depending on which platform is used
… sometimes the meta-model is given and it's only a matter of the headings
… in other cases there is more freedom
… The question is if we want to model it in detail or just link the dataset description to a profile
… we need to think of that first and then we can decide on how to describe it (in DCAT or outside)
… Not sure if there is a use case for that

Jaroslav_Pullmann: finds it important to find the parts that refer to a world context (feature of interest etc.).
… other parts of semantic expression is the data itself
… perhaps we should split: real world entities vs. data gained
… we should be clear about the outcome of the requirement
… suggests to split off the two meanings of semantics

<roba> should we remove word "semantics" from this definition - make it "formal description of aspects" ?

Makx: from practical experience: they made a specific AP for statistical data in order to give semantics to those dimensions
… but often it's a matter of domain specific dimensions and then it might not make sense to have common properties

Jaroslav_Pullmann:
… agrees with Makx. Asks Makx to add a use case for features of interest. This was discussed at the Oxfort F2F (real world objects)

Jaroslav_Pullmann: sees a twofold topic:
… 1) what is considered the source of data
… 2) how to interpret that
… the first is a contextual reference (real world domain entity)
… could be an invoice, a shipment

Makx: has nothing to do with the comment.
… some data might be in Data qb, others in SDMX

<Caroline_> s/ack Jaroslav_Pullmann

Makx: if you have particular domains with particular data then that domain might have a particular
… interest to expose certain internals of their data at the dataset level

Jaroslav_Pullmann: only referred to structured data
… felt that there are many references in the UC that refer to the real world
… but nothing talking about the data points

(hope I got that right...)

roba: fine grained detail in observations and measurements often gets blurred
… end users have different perspectives (related but not the same)
… suggests to push things to profiles and then have best practices
… for domain-specific applications
… thinks there will be specialisations of general profiles
… suggests to remove the word "semantics" and replace with "specific aspects"

<alejandra> or "formal aspects"

kcoyle: sounds as if Jaroslav_Pullmann talks about things that help with discovery, suggests to provide a UC for that since there is none

Jaroslav_Pullmann: will do , agrees that that's the aspect

kcoyle: suggests that we consider this as something to relegate to profiles and to use roba's language for that
… and assumes that the profile will be linked to the dataset using a to-be-defined property

<trackbot> Error finding 'on'. You can review and register nicknames at <https://‌www.w3.org/‌2017/‌dxwg/‌track/‌users>.

Action: Jaroslav_Pullmann to create use case relating to real world entities as source of data

<trackbot> Created ACTION-50 - Create use case relating to real world entities as source of data [on Jaroslav Pullmann - due 2017-10-30].

roba: there are a number of specialised requirements here so there are some different examples of the same requirement

kcoyle: is there a link from the req to the UC?

roba: yes, and the forward-links from UC to requirement isn't done yet

Jaroslav_Pullmann: the forward-links from UC to req are normally created automatically
… in order to have just one source to maintain

<kcoyle> PROPOSED: accept 6.13 using Rob's suggest rewording; associate with profiles

<PWinstanley> can hear you caroline

<riccardoAlbertoni> \me yes

<Caroline_> can you hear me?

<kcoyle> suggests to remove the word "semantics" and replace with "specific aspects" [07:46] <alejandra> or "formal aspects"

Jaroslav_Pullmann: wants to mention the aspect of interpretation of data

kcoyle: the wording should be specific aspects or formal aspects

roba: should be fomalisations of descriptions of data (for discovery or whatever)

Jaroslav_Pullmann: not very happy with that (yet)

<alejandra> +1 to Jaroslav_Pullmann I think it would still be too vague

Jaroslav_Pullmann: thinks we should express what the intention of the description is and put that in the requirement title

<Caroline_> PROPOSED: accept 6.13 using Rob's suggest rewording; associate with profiles

<kcoyle> +1

<Caroline_> +1

<Jaroslav_Pullmann> +1

<DaveBrowning> +1

<roba> +1

<Stijn_Goedertier_AIV> +1

<dsr> +1

<riccardoAlbertoni> +1

<PWinstanley> +1

<alejandra> I think the proposal is not too clear

Resolved: accept 6.13 using Rob's suggest rewording; associate with profiles

<Makx> +1

<AndreaPerego> +1 - but agree with alejandra that it needs to be clarified

alejandra: wants to vote on the reworded requirement instead

kcoyle: Ixchel has already made a similar proposal

<alejandra> this is the text kcoyle said

<alejandra> "6.13.1 Provide recommendations and mechanisms for data providers to describe datasets with fine grained semantics (e.g. instrument/sensor used, spatial feature, observable property, quantity kind)."

<alejandra> we should have discussed that text

<Caroline_> https://‌www.w3.org/‌2017/‌dxwg/‌wiki/‌F2f2

Action: Ixchel to update text in 6.13

<trackbot> Created ACTION-51 - Update text in 6.13 [on Ixchel Faniel - due 2017-10-30].

<riccardoAlbertoni> Thanks you bye

<AndreaPerego> Thanks, and bye!

<PWinstanley> bye

<Makx> bye

<Jaroslav_Pullmann> bye

<alejandra> thanks, and bye!

Caroline_: reminds all to add F2F attendance to the wiki

bye

– DRAFT –
Dataset Exchange Working Group Teleconference

23 October 2017

Meeting Minutes

Meeting time post-daylight savings

Open Actions

UCR

Summary of Action Items

Summary of Resolutions

Diagnostics