[data-protocols] Looking inside tables

[Omar Benjelloun]

Hi Xavier,

Thanks for your feedback! I agree with you that the Look Inside proposal
tries to address more general tables, not just statistical tables, and
shares with Json-Stat the goal of simplicity.

You're also right that this approach doesn't work for many tables on the
Web (and CSVs for that matter). This trade-off is what allows the language
to remain simple.

I believe "exotic" tables can be addressed by creating ETL/ transformation
scripts to normalize them, and ideally open sourcing these transformation
scripts and publishing them in repositories like github.org/datasets.

I think an important aspect of this proposal is to use vocabularies like
schema.org or other RDF vocabularies to define the schema / semantics of
the data. This should help make datasets more interoperable, and
discoverable, which should increase their usefulness.

Cheers,
-Omar




On Fri, Aug 16, 2013 at 3:32 PM, Xavier Badosa <xbadosa at gmail.com> wrote:

> Very interesting.
>
> The idea is to provide a mechanism to map CSV or HTML tables into semantic
>> vocabularies
>
>
> I'm not sure CSV and HTML tables can be considered together.
>
> This proposal seems to seek a similar goal as the StatisticalTable schema
>
> http://json-stat.org/schema/
>
> The StatisticalTable schema, though, focuses on a specific type of table
> (the statistical table, where the cell content is a measure). Both try to
> avoid redundancy and verbosity assigning properties directly to columns.
> The StatisticalTable schema allows assigning such properties also to rows.
> This seems natural in the case of HTML tables (and presentation tables in
> general, where choosing between rows or columns seems a layout decision).
> On the other hand, considering columns different than rows is the CSV (or
> the relational databases) approach.
>
> In summary, the proposal probably provides a mechanism to map CSV tables,
> but, if I understand it correctly, many existing HTML tables on the web
> won't be able to benefit from such enriched markup.
>
> Am I right?
>
> X.
>
>
> On Fri, Aug 16, 2013 at 4:47 PM, Omar Benjelloun <benjello at google.com>wrote:
>
>> Hi all,
>>
>> You may be interested in the proposal I posted to the
>> public-vocabs at w3.org list.
>>
>> http://www.w3.org/wiki/WebSchemas/LookInside
>>
>> The idea is to provide a mechanism to map CSV or HTML tables into
>> semantic vocabularies (e.g. from schema.org) to enable a semantic
>> understanding of their actual contents.
>>
>> The examples in the above description focus on simple tables / existing
>> schema.org vocabularies, but I believe this approach can be useful in
>> many open data scenarios, e.g. to enable discovery, composition and
>> visualization of datasets.
>>
>> Your thoughts / comments are welcome!
>>
>> Best,
>> -Omar
>>
>> --
>> Omar Benjelloun | benjello at google.com | (415) 845-8516
>>
>> _______________________________________________
>> data-protocols mailing list
>> data-protocols at lists.okfn.org
>> http://lists.okfn.org/mailman/listinfo/data-protocols
>> Unsubscribe: http://lists.okfn.org/mailman/options/data-protocols
>>
>>
>


-- 
Omar Benjelloun | benjello at google.com | (415) 845-8516
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/data-protocols/attachments/20130816/97304cb0/attachment.htm>