Re: Feedback on LD glossary

Hi Bernadette,

On 19/03/13 21:27, Bernadette Hyland wrote:
> Hi Dave,
> Guess you're back from holiday ... ;-)

Sort of, still recovering (from bug, not injury).

>> > 2. 5 Star Linked Data Diagram
>>
>> Drop this entry.
>
> I think it adds value & is line with discussions the GLD WG has had to
> provide diagrams & collateral that help describe the value of LOD for
> open gov't content.  I've emailed Michael H. to confirm however that it
> is published with a suitable open content license as that is a must have.

If the diagram adds value then perhaps make it part of the entry on 5 
star, not a separate entry in its own right.

>> > 24. Data Warehouse
>>
>> Hmmm. Possible rewrite:
>>
>> [[[
>> A Data Warehouse is one approach to data integration in which data
>> from various operational data systems is extracted, cleaned,
>> transformed and copied to a centralized repository. The centralized
>> repository can then be used for data mining or answering analytical
>> queries.
>> ]]]
>>
>> That rewrite misses out the red-rag-to-a-bull comment on how Linked
>> Data is an alternative. The story there is a lot more complex than the
>> existing entry suggests. If you really want something to about the
>> relationship to linked data then that will take rather more work to
>> phrase just right.
>
> Thanks for the proposed rewrite.  Yes, the story of how LD compares is
> more complex but something (accurate but brief) is better than nothing
> IMO.  How about:
>
> <h4>Data Warehouse</h4>
> A data warehouse is one approach to data integration in which data from
> various operational data systems is extracted, cleaned, transformed and
> copied to a centralized repository. The centralized repository can then
> be used for data mining or answering analytical queries.  By contrast,
> Linked Data <em>assumes and accounts</em> for a <em>distributed
> approach</em> using HTTP URIs to describe and access information
> resources.  A Linked Data approach is seen as an valid alternative to
> the centralized data warehouse approach especially when integrating open
> government datasets.
>
> We remain open to editing...

Well that omits the IMHO equally valid option of materializing your 
integrated data as a big pile of RDF. The implementation doesn't *have* 
to be distributed and often centralized is the only option that gives 
sufficient performance.

However, saying that well would require an entire essay so I'll accept 
your version.

>> > 33. Dublin Core Element Set
>>
>> I tend to think of DC Elements as referring to 1.0 and DC Terms to
>> 1.1. Certainly the page you link to is called "dmci-terms".
>>
>> Suggest s/Element Set/Metadata Terms/ in both title and body.
>
> Good catch.  How about we just stick with the core 15 terms widely used.
>   Are you OK with this?

I would prefer to refer to "Dublin Core Metadata Terms". There has been 
a preference for using DCT rather than old DC elements in all the public 
sector I've been involved in for some years.

However, I'm not a DCMI person, presumably there's others in the group 
better placed to advise on correct terminology.

>> > 39. Hash URI Strategy
>>
>> Urgh. That one needs a rewrite (and to be paired with one slash URIs).
>> Run out of time to suggest something now ...
>
> How about we remove it.  I don't hear many open gov data people chatting
> in the halls about hash URI strategies or loosing sleep over it.  Removed.

OK, in my follow up message there is a draft. Feel free to use that or 
leave it removed.

Cheers,
Dave

Received on Tuesday, 19 March 2013 22:56:13 UTC