F2F Agenda 2015-02

CSVW WG F2F Meeting, 12-13th February 2015

123 Buckingham Palace Road (c/o Google UK), Victoria, London, UK

TODO:

• Meeting attendance registration form (set date for close of registration)
• Hotel recommendations would be useful - please add if you have suggestions

This is the Meeting Page for the CSV on the Web WG (CSVW) 12-13th February 2015 F2F meeting in London, UK.

This Meeting Page, in particular the Agenda section, is a work in progress and may change

If you are attending, please let us know via the Attendee List. If not, please write your name there under Regrets.

Meeting Logistics

• Meeting Host = Google (contact: Dan Brickley, danbri@google.com)
• IRC channel = #csvw; irc.w3.org:6665
• Dates = February 12-13 2015
• Time = See below; all times are the local time zone
• Location = c/o Google UK, 123 Buckingham Palace Road, London, UK. SW1W 9SH.
• Meeting Room = To be confirmed.
• Voice Conference Bridge: W3C's Zakim Bridge will be used if/when needed. +1.617.761.6200; see also Zakim SIP Bridge; PIN = 2789# (CSVW)
  • Within the meeting room, to connect to Zakim enter the following in IRC: zakim, call @TBD

Attendee List

Attendees:

PLEASE SIGN UP HERE BEFORE JAN 25th (just write your name, or link to a mailing list post).

Meeting room space is limited so this information is essential for planning.

If you have physical accessibility or diet considerations that you think might be useful to know about, please communicate those with Dan directly. We'll eat at a Google cafe which should cover most common diet situations, but we can investigate for allergies and other constraints.

• Dan Brickley
• Ivan Herman
• Jeni Tennison
• Gregg Kellogg
• Jürgen Umbrich (WU Vienna)
• Jeremy Tandy
• Phil Archer
• Rufus Pollock (maybe)
• Davide Ceolin
• your name here...
• ...

Regrets:

(please say if you won't attend, so we aren't left wondering)

• Bill Ingram
• Andy Seaborne
• ...

Please note that we have relatively small rooms booked. The WG chairs and staff contacts believe they are adequate

Logistics and Hotels

This 2-day meeting is hosted at one of Google's London offices: 123 Buckingham Palace Road. It is next door to Victoria Station.

Please see Dan's mail for other logistics info.

Please be aware that there are several other Google addresses in London, including another office nearby on the same road.

Two (different) rooms are booked: a 12 person meeting room (day 1) + an 8 person room (day 2). These were the largest available; Dan is investigating possible improvements.

At this time we have no hotel recommendations. Please share any useful links and information here. Local host contact is Dan Brickley.

See also:

• Google map for location
• Trip Advisor hotels near Victoria
• Ivan has suggested a shortlist

Agenda

We have no formal agenda yet, but the overall focus will be on finalizing our specifications, and will be driven by our github issue list.

See WG homepage for Working Drafts and Charter

Thursday 12th February

Chair: @@@

Meeting room: @@@

NOTE: Agenda topics below are a skeleton based on the last F2F. They need discussion. This is just a shell to propose some rough timings.

09:00 - 10:30 Overview

• Introductions
• Review of charter & scope
• Review of what we've done
  • Model spec
  • Metadata spec
  • Conversion to RDF
  • Conversion to JSON
  • Test suite
• Review of future timeline & constraints
• Goals for the F2F
• Agenda review & adjustment

10:45 - 12:15 Parsing CSV & other tabular data

• Review of data model
• Parsing CSV
  • Documenting the processing steps for metadata
  • Locating data regions within a CSV file
  • Should we check integrity of CSV files in some way
  • Fall-back value of the `name` property?
  • Should there be limitations on the syntax of column names?
  • How to determine language encoding
• Handling of formats other than CSV

13:00 - 14:30 Metadata Discovery & Merge

• Metadata discovery
  • Locating additional metadata when originally starting from a metadata document
  • Security considerations of modifying path for metadata
• Metadata merge algorithm
  • Need default metadata
  • Slight confusion with arrays an atomic values in the proposed metadata merge scheme
  • Property value of Natural Language property should return all language values

14:45 - 16:15 File vs Model

• CSV file != table model
• Where does row numbering begin?
• Simplify the definition of the tabular data model?
• There should be a "Fragment Model" section in the Data Model document

16:30 - 18:00 Metadata / Annotation handling in conversion

• Lack of example for canonical mapping use case
• Metadata
  • Value space of Common Properties
  • What should the name of the property be that relates rows to the table?
  • Should we remove column metadata in RDF mapping
  • Should the "@type" information be reflected in the generated JSON/RDF?
  • Using csvw:Table in RDF output conflates a metadata definition table definition
  • Should the table entity in the RDF mapping of core tabular data be explicitly identified?
  • Is the DCAT block useful in the RDF output.
  • Should the Dialect description be related to the dcat:Distribution of the Table?
  • Prov roles
• Annotations
  • Exact handling of annotations

Friday 13th February

Meeting room: @@@

Chair: @@@

09:00 - 10:30 Primary Keys & Cross References

• Limit foreign key cross references to the same batch of processed resources
• Should primary keys be skipped from cell level triple (or k/v pairs) generation?
• Alternate structure in JSON mapping output graph if Primary Key is specified
• Making `schema` property mandatory for table description objects & explicit identification of schemas
• Composite primary keys and foreign key references

10:45 - 12:15 URLs in Metadata

• URL expansion
  • How simple is simple mapping?
  • Expansion of prefixes within URI templates
  • What is default value if @base is not defined in the metadata description
  • Should we use a JSON-LD reference for relative URLs?
• Extension Conversions
  • Possible error in "optional properties" for Template Specifications: source

13:00 - 14:30 Conversion Details

• Conversion issues
  • Suppression of columns in mapping
  • Do we need "phantom" columns, i.e., columns with their own and separate 'aboutUrl' value?
  • How should class level qualified properties be transformed to JSON
  • Rich Column Classes / Types (@type / @datatype on column)
• Datatypes
  • Make the datatype mapping more precise
• Lists
  • Should there be an option for unordered lists in RDF mapping?
• Terse description of conversion output

14:45 - 16:15 Overflow Time

• Conflation of metadata descriptions & table descriptions
• Relationship between table group, table and schema
• Metadata merge
  • Need default metadata
• Conversion to RDF
  • Should the "@type" information be reflected in the generated JSON/RDF?
  • Is the DCAT block useful in the RDF output.
  • Should the Dialect description be related to the dcat:Distribution of the Table?
  • Prov roles
• Define set of examples to use through documents
• Relationship in RDF output of conversion between csvw:Table and the entities generated from a row
• Lists
  • Should there be an option for unordered lists in RDF mapping?

16:30 - 17:15 Wrap-up

• Action review
• Timeline plan
• AOB