CSV on the Web Working Group Teleconference

<trackbot> Date: 02 April 2014

<danbri> thanks, AndyS!

<danbri> [high pitched whooshing noise]

<DavideCeolin> muted, can you still hear the noise?

can try...

<scribe> scribe: Axel Polleres

<scribe> scribenick: AxelPolleres

<phila> scribe: AxelPolleres

<phila> scribeNick: AxelPolleres

<phila> Guest: PhilA

<danbri> agenda https://www.w3.org/2013/csvw/wiki/Meeting_Agenda_2014-04-02

danbri: approve minutes from last meeting http://www.w3.org/2014/03/26-csvw-minutes.html

<AndyS> Minutes looks OK

<danbri> http://www.w3.org/2014/03/26-csvw-minutes.html looks ok to me

RESOLUTION: minutes from last meeting http://www.w3.org/2014/03/26-csvw-minutes.html approved

<jtandy> ?

danbri: 2 FPWDs shipped last week... any feedback already?

<AndyS> Don't see anything on public-csv-wg-comments@w3.org

danbri: any news from the editors?

jeremy: none

danbri: any news from the best practices WG meeting?

phil: reporting from DWBP WG

<danbri> http://www.w3.org/2013/dwbp/wiki/Main_Page

(... as team contact)

<ericstephan> Great to have you on the call this week and next phila!

phil: ... closely related to CSVW
... develop voacularies on data usage, quality, usage best practices on existing vocabs, e.g. PROV.

<phila> http://www.w3.org/blog/data/

<danbri> ACTION: danbri write (or find writer for) data activity blog re new WDs [recorded in http://www.w3.org/2014/04/02-csvw-minutes.html#action01]

<trackbot> Error creating an ACTION: could not connect to Tracker. Please mail <sysreq@w3.org> with details about what happened.

<danbri> unmute me

phil: ... somone from this group should draft something for the data activity blog on the CSVW group.

eric: phil has brought up a use case which I'd like to propose tro the WG. table from PDF docs.
... once it is tabular data, it is relevant. interesting use case that can go between the two groups.

danbri: use case doc is the ideal interface between the group.

phila: most people in DWBP WG haven't read it yet, needs to be done. I have another use case...

<jtandy> +1 to phils use case

phila: person a creates tabular date w/o metadata, person b creates metadata and wants to link it to tabular data from person a.

<AndyS> phila: third party metadata UC

<danbri> ACTION: chase phila for his usecase in which a party provides metadata for another's csv [recorded in http://www.w3.org/2014/04/02-csvw-minutes.html#action02]

<trackbot> Error creating an ACTION: could not connect to Tracker. Please mail <sysreq@w3.org> with details about what happened.

<phila> ACTION: dan to chase phila for his usecase in which a party provides metadata for another's csv [recorded in http://www.w3.org/2014/04/02-csvw-minutes.html#action03]

<trackbot> Error creating an ACTION: could not connect to Tracker. Please mail <sysreq@w3.org> with details about what happened.

<danbri> (trackbot issue is known/reported. … we have chatlogs for minutes)

Andy: request feedback on expectations on how standardised mapping from CSV can be, can we use input from the other WG?

CSV2RDF

danbri: andy & greg report back?

andy: greg and i will have a separate telco, to spare general telco time, will come up with rough drafts.

danbri: I support that.
... stay as transparent as possible, share minurs, call times

<AndyS> 5pm UK / 9am PDT / 6pm CEDT / 12:00 EDT

<AndyS> http://w3c.github.io/csvw/csv2rdf/#example-1

<jtandy> tuesday

<danbri> tues

<AndyS> Tuesday

<danbri> :)

spinoff-telco will be Tue, 5pm UK / 9am PDT / 6pm CEDT / 12:00 EDT

andy: column mapping, where each cell in a row becomes a particular value, guessing datatypes, is all in scope... where we'd leave strictly tabular data is starts getting more difficult.

greg: generally, uninformed mapping that should work without metadata describing the content

<AndyS> https://github.com/w3c/csvw/tree/gh-pages/tests

greg: ... (continues on testcases)

<danbri> which are the rdf ones?

<AndyS> .ttl

greg: json doesn't imply order on elements

<AndyS> http://w3c.github.io/csvw/csv2rdf/#example-1

jeremy: will alternate telcos also have Zakim, minutes, etc.?

andy: we need someone programming Zakim

danbri: adhoc calls?

phila: adhoc can be scheduled, it's simple, if regular calls, we shall set it up.
... (ivan or phil)

phil explains how to setup adhoc calls...

phil: note we have a limited number of lines.
... you can still use IRC even if you can't use zakim.

greg: can we use rrsagent, trackbot?

<Zakim> danbri, you wanted to discuss r2rml as candidate 'how to do complex mappings' option

phila: trackbot not working at the moment

danbri: have been reporting on that.

jeremy: mapping predicates on column headings, sometimes it's more complicated, e.g. if I want to use units (QUDT)
... I put that as a requirement in the UC document.

andy: if we fulfill every requirement, we are getting in the complicated space
... mapping not completely uninformed, may use metadata

danbri: where to draw the line?

andy: if it starts to getting to conditionals, that's where it starts to fell complicated (going into programming languages)
... we need concrete cases

<jtandy> the requirement for LDPath "predicates" for column mapping is provided as a note in http://w3c.github.io/csvw/use-cases-and-requirements/#R-SemanticTypeDefinition

andy: "this column is dc:title" is quite within column mapping

<danbri> "supporting automated recognition of semantic type, typically expressed for each column ... further description to be added"

andy: would be helpful to have jeremy's example on RDF output expected.

jeremy: agree that the example I am talking about needs multiple jumps beyond uninformed mappings

danbri: r2rml allows SQL for more complex mappings
... we should look at real world r2rml examples, have been looking at musicbrainz examples.

<jtandy> ACTION: jtandy to provide detailed description of the LDPath as column mapping requirement, including the anticipated RDF [recorded in http://www.w3.org/2014/04/02-csvw-minutes.html#action04]

<trackbot> Created ACTION-11 - Provide detailed description of the ldpath as column mapping requirement, including the anticipated rdf [on Jeremy Tandy - due 2014-04-09].

danbri: r2rml looks like a good handle.

andy: if you want to use complicated stuff, load your csv into SQL and use r2rml?
... full sql is turing-complete.

<phila> "expression": "'/id/'+ toLowercase(replace(replace(value, /(ministry|office|department|managing|directorate|unit\\s(general|authority)?).*/i, \"$1\"), /\\s+/, '')) + '/' + value.urlify()",

<AndyS> andy: due recent standards additions.

phila: (on his irc example) does that count as complex?

<phila> https://github.com/OpenRefine/OpenRefine/wiki/GREL-String-Functions

phila: is that in scope?

andy: I think defining a library for that would take quite some time... that would be a big WG decision (more on time it takes to do the work than on technical issues)

phila: many mappings quickly need string functions

<Zakim> danbri, you wanted to ask if there are functions we could refer to from other specs eg sparql?

andy: depends on the library, also from what programming environment you are coming from.

danbri: sparql functions and operators?

andy: sparql ones are all from XQuery/XPath functions, differences in regex replacement

danbri: maybe postpone that to version 1.1

andy: idea would be to publish in 2 phases, starting with a basic extensible version

danbri: we can screw up in both directions, making it too toy-like to be useful, or too complex to implement

greg: csv to jjson is interesting for loading it into tools, putting too much into it, doesn't fulfill the purpose...

<danbri> AndyS, it is in fact a fair question whether we go through and mark UCs as 'accepted'

<danbri> or maybe as 'implemented by x' where x is a piece of spec

greg: case conversions, modifiers, chaining operations go to far? JSON-

l

scribe: LD suffers from that a bit. too many new use cases coming in.

danbri: actions for RDF work for next week?

andy: will try to sketch something on requirements

danbri: anything on i8n?
... or should we move on to json?

<AndyS> AndyS: need more people involved in JSON conversion.

jeremy: would be interested in json, but over the next week other priorities

<jtandy> sadly - next few weeks ...

danbri: adhoc f2f meetings in london possible with some members around
... AOB?
... adjourned.