CSV on the Web Working Group Teleconference

07 May 2014


See also: IRC log


Jeremy Tandy (jtandy), Phil Archer (phila), Andy Seaborne (AndyS), Dan Brickley (danbri), Ivan Herman (Ivan), Yakov Shafranovich (yakovsh), Eric Stephan (ericstephan), David Ceolin (DavideCeolin), Jürgen Umbrich (jumbrich), Jeni Tennison (JeniT), Alfonso Noriaga (fonso)
Dan Brickley
Phil Archer


<danbri> last week's meeting record: http://lists.w3.org/Archives/Public/public-csv-wg/2014Apr/0123.html

<AndyS> http://www.w3.org/2014/04/23-csvw-minutes.html

<danbri> 30th april minutes: http://lists.w3.org/Archives/Public/public-csv-wg/2014Apr/0123.html

Previous week's minutes are at http://www.w3.org/2014/04/23-csvw-minutes.html

i.e. it was on the 23rd, not the 24th

<AndyS> +1 to both

PROPOSED: Accept previous two week's minutes

<ericstephan> +1

RESOLUTION: Accept previous two week's minutes http://lists.w3.org/Archives/Public/public-csv-wg/2014Apr/0123.html and http://www.w3.org/2014/04/23-csvw-minutes.html

<scribe> chair: danbri

<scribe> scribe: phila

<scribe> scribeNick: phila

danbri: Introduces agenda

Use Cases and Requirements doc

danbri: Where are we?

<ivan> UCR Editors' draft

ericstephan: Folks have been pretty swamped but we've managed to add a few more use cases
... one came from Tim Robinson this morning
... and one to come from Phil

phila: (I added it this morning too)

ericstephan: We have to firm up the categorisation

<jtandy> do we need to flesh out the requirements themselves?

danbri: So do you have what you need for a next publication?

ericstephan: It depends on the time all of us have this week
... I'm going to be fairly busy but I can spend some time on it

jtandy: I've not been able to pay much attention so far this week but I should be able to support Eric this week

DavideCeolin: There are a few things to discuss. I sent an e-mail about that
... I have tried to classify the requirements following the last meeting. I don't know if the classification is OK or not
... also I was doubtful about the annotation req. Some reqs could be sub-reqs of that

<DavideCeolin> http://lists.w3.org/Archives/Public/public-csv-wg/2014May/0003.html

<danbri> thx


phila: I added my use case. I'm now checking with the people it talks about that it is accurate

ivan: I think the new use cases lack requirements
... Publishing in 2 weeks is not sacrosanct
... I'd rather that the requirement table and put it in the doc. If that takes > 2 weeks, so be it
... what's on the wiki is a significant step forwards

ericstephan: I would prefer moving without doing what Ivan just suggested

<jtandy> +1 to taking more time to get the requirements correct

<DavideCeolin> +1

<jtandy> ... and taking more time as necesssary

ericstephan: I think we can try to get things done... I'd prefer if we had a bit more time (Scribe correct, i.e. Eric agreed with Ivan)

<danbri> [ah, 'moving out' was wrt to timescale, not location of text]

ivan: There is no requirement to publish on 2 - 3 weeks

JeniT: I think this doc is really good, the classification is excellent the reqs (section 4) ...
... section 4.2... I suggest putting all the reqs into accepted for what we want to publish next
... The 2 - 3 weeks deadline was arbitrary because I think deadlines help to motivate people
... if it takes longer, then OK
... but it's also good to publish heartbeat docs so people can track even small changes
... Re the issue of annotation and supplementary info reqs...
... There were lots of sub reqs and it's not clear whether we need some of the high level reqs

JeniT: I think we're looking at human-readable annotations

<jtandy> +1

<JeniT> http://w3c.github.io/csvw/use-cases-and-requirements/index.html#R-AnnotationAndSupplementaryInfo

^^ That's the specific req under discussion

JeniT: There are lots of reqs, sub reqs around that

<jtandy> I agree that the "annotation and supplementary info" requirement should not overlap with the scope of the other requirements - therefore focus on the human readable annotation

RDF Mapping

danbri: This will be a FPWD of the RDF Mapping?

<ivan> Generating RDF from Tabular Data on the Web

AndyS: Only if you want a very rough and ready document
... There's a lot to be thought through to align the docs

danbri: How would an extra week help?

AndyS: I don't think it will make a stunning difference

danbri: Any particular reason for publishing everything in coordinated waves?

ivan: No, we can do what we like in that regard
... What bothers me about publishing this doc now... we had a discussion a few weeks ago on the general principles of conversation of CSV
... I think what the doc has is more or less in line with that. But I would like those discussions documented and published somewhere
... just for the public view of things

JeniT: Which doc would that fit in best?

ivan: I was thinking the same thing but don't have an answer

JeniT: There might need to be a general conversion doc?

ivan: Might it fit in the structure of the model doc?

JeniT: I can look

ivan: I don't mind. I just think the principles should be included somewhere

AndyS: Could you take a go at writing a couple of paras in an e-mail then it can at least go in the RDF doc. It might get moved later

ivan: It's a possibility, but if we do a JSON doc at some point then we'll have to sync those texts

AndyS: That might be the triggerto get it out of the RDf doc
... Let's get the text out now and worry about where it ends up later

<JeniT> +1

danbri: Jeni just circulated a draft

<danbri> ACTION: danbri supply a json-ld sentence for csv2rdf doc [recorded in http://www.w3.org/2014/05/07-csvw-minutes.html#action01]

<trackbot> Created ACTION-14 - Supply a json-ld sentence for csv2rdf doc [on Dan Brickley - due 2014-05-14].

AndyS: I thought Jeni's message is about an RDF form of the metadata doc

danbri: The idea was to take Jeni's sentiment and add it to the doc

JeniT: Not sure how that's relevant?

danbri: I'm just thinking about someone coming in to this and seeing RDF in 2 places

ivan: There is the issue of how the three docs work together

<AndyS> The CSV2RDF doc will refer to metadat doc and (eventually) align.

ivan: When I first saw the metadata doc I was a bit confused about the relationships. It needs clarity, that's all

danbri: Any more to say on RDF generation?

AndyS: If you can send the mail, Ivan, I'll add it to the doc

<scribe> ACTION: Ivan to send 2 paras on format conversion in an e-mail that Andy will incorporate in the mapping Doc [recorded in http://www.w3.org/2014/05/07-csvw-minutes.html#action02]

<trackbot> Created ACTION-15 - Send 2 paras on format conversion in an e-mail that andy will incorporate in the mapping doc [on Ivan Herman - due 2014-05-14].

ivan: We haven't got any use cases for the XML conversion?

JeniT: No, but we did just have a use case with XML metadata

ivan: It would be good to track that

JeniT: I agree
... I have taken CSV files and converted them to XML but there was more to it than that

Model for Tabular Data

danbri: You had a proposal, Jeni that everyone agreed with

<ivan> editor's copy

JeniT: Yes, the proposal was to not worry about embedding any metadata in the CSV other than the column headings
... lots of +1s to that

<jtandy> +1

<ericstephan> +1

danbri: So we're saying nothing, not that you can't do it?

JeniT: We're saying for our WG that, at least for now, we're focussing on CSV with no extra header lines with reserved character starts

danbri: Yes but people *are* putting extra headers in

JeniT: Yes, lots of examples of that - random ways - we need to scope that out for now I think

danbri: But we still need to point to the metadata file

JeniT: Yes
... We have resolved to defer any consideration of embedding metadata in CSV docs with the exception of column headings


JeniT: There are a few other issues... but I'm generally happy for them to stay in the file and for us just to have a heartbeat publication, prob in line with otehr docs

danbri: Any more on this doc?

ivan: I want to make sure that this is reflectied - the RtL issue? We have the use case
... that's a new req that should be documented that may have an effect on the parsing algorithm

JeniT: I'm happy to add something about that
... prob in the parsing bit which is informative

ivan: I tried to get some feedback from CN and JP folks to see if vertical writing might be relevant too
... it seems that no, they tend to work in rows (LtR)

<jtandy> re JP and CN - that's good to hear

ivan: but Arabic and Hebrew def go RtL

danbri: Do you have all you need?

JeniT: Yep

metadata document

<danbri> http://lists.w3.org/Archives/Public/public-csv-wg/2014May/0017.html

<danbri> http://w3c.github.io/csvw/metadata/

JeniT: Last week we had a rough version. This week's is slightly less rough

<danbri> "Validation and conversion of tabular data on the web requires additional metadata that describes how the data should be interpreted. This document defines a vocabulary for metadata that annotates tabular data. This can be used to provide metadata at various levels, from collections of data from CSV documents and how they relate to each other down to individual cells within a table."

JeniT: I have discussed with Rufus how to structure this
... One was whether to take it vocab first and then to JSON or from format to vocab - Rufus said the latter
... the other things was orienting it around using JSON-LD to map the data format
... also decided to scope the simplest possible case which is a single CSV file described by a single metadata file
... that leaves out packages of CSV files
... and several CSV fails that share the same schema
... what I've done today is to see those divisions in the format of the doc

<JeniT> http://w3c.github.io/csvw/metadata/#tables

JeniT: looking at aligning that with existing RDF vocabs using JSON-LD
... using @ to point to the CSV file
... we can probably guess a rough version together over the next couple of weeks or a more formed version in 4 weeks
... So I'd welcome any comments on approach and direction. It means that we depart quite a bit from teh original data package format

ivan: So do we want at some point to go through the various issues before publication?

JeniT: I think it's OK to have some issues in teh doc but it would be good to go through them
... anything up to section 2,2 basically

ivan: I think it would be good if this doc included a reference/example to the RDF conversion doc
... AIUI if we use some sort of template for the RDF conversion then it should be clear where it goes in this metadata doc

<jtandy> agreed - it's not clear how the RDF conversion templates fit into the metadata doc

ivan: but it's not clear to me yet

JeniT: Noted
... The format is very much taken from the data packages spec which doesn't include anything about conversion. I think that's a feature of the doc being young

danbri: Does it have any relationship to the UCR (explicitly)

JeniT: Not explicitly. Reqs are quite high level. e.g. 'Validation' - meaning what kind of validation?

jtandy: I just wanted to say... following on from the weather obs example I put on github, I'll try and apply the metadata doc and if broken will feed back

JeniT: That would be v helpful - but give it a week first

<danbri> (phil/ivan, where are we w.r.t. idea of rdf validation wg at w3c?)

<Zakim> phila, you wanted to talk about validation

discussion of poss usefulness of proposed RDF validation

JeniT: validation of the JSON might be useful for people wanting to use a JSON schema

<danbri> ack me?

danbri: The RDF validation might take some pressure off us but we can't get wawy without any validation
... Anymore on the metadata vocab doc jeni?


danbri: Any upcoming events?

<ericstephan> +q

<danbri> http://2014.eswc-conferences.org/

<JeniT> http://www.thinkingdigital.co.uk/

<JeniT> http://csvconf.com/ !!!!!

<DavideCeolin> possibly, but not 100% sure I'll be at ESWC

ericstephan: I'll be at hte data Reproducabilty workhop in Seattle tomorrow. Focused on the biomedical community. Very spreadsheet-centric. Lots of discussion there on tools and standards

<danbri> this? http://escience.washington.edu/event/first-reproducibility-workshop

<danbri> ESWC? seems not.

<ericstephan> yes

JeniT: There's a CSV Conf!
... the first international CSV conference in Berlin

phila: notes "A conference for data makers. Part of the week long OKFestival"

<danbri> part of http://2014.okfestival.org/ (so Rufus might know more)

ivan: Is it important for us to be there and for them to know about our work

<danbri> jeni planning to go

JeniT: I'm planning on going and hoping to give a presentation

ivan: Great

danbri: I'd like to go if I can

<ericstephan> Thats our slogan CSV is relaxing

<yakovsh> there is going to be an ietf meeting in canada in july but its probably not relevant for our work : http://www.ietf.org/meeting/90/index.html

meeting adjourned

Summary of Action Items

[NEW] ACTION: danbri supply a json-ld sentence for csv2rdf doc [recorded in http://www.w3.org/2014/05/07-csvw-minutes.html#action01]
[NEW] ACTION: Ivan to send 2 paras on format conversion in an e-mail that Andy will incorporate in the mapping Doc [recorded in http://www.w3.org/2014/05/07-csvw-minutes.html#action02]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2014-05-07 13:06:36 $