Re: [PROV-AQ] Multiple anchors and provenance-URIs

Hi Tom,

Short answer: the first HTTP example is OK, and cannot be exactly replicated in 
HTML.

It's a known wrinkle, and has also been noted in review comments.

The choice we made was to trade maximization of use of existing capabilities 
over consistency, so as you say there is a difference in selectivity of Link: 
headers vs HTML <link> elements.

My position is that multiple hasProvenance links is probably an edge case, and I 
don't want to get too drawn into optimizing for it.  Accordingly, the general 
position is that you have one or more provenance URIs and one or more anchors: 
you can then look for references to any of the anchors in any of the provenance 
records.  This may be slightly inefficient but it cannot yield incorrect results 
if the provenance data is correct.

The HTTP case can be optimized because you are given anchors values on a 
per-provenenance-record basis.  But it's just that, an optimization, no more.

This is intended to be covered by this text:
[[
The mechanisms described here are intended to allow a provider to supply 
information that may assist a client in finding related provenance records. The 
provenance records should themselves explicitly identify the target resources 
they describe. While a provider should avoid giving spurious information, there 
are no fixed semantics, particularly when multiple resources are indicated, and 
a client should not assume that a specific given provenance-URI will yield 
information about a specific given target-URI. In the general case, a client 
presented with multiple provenance-URIs and multiple target-URIs should look at 
all of the provenance-URIs for information about any or all of the target-URIs.
]]
-- 
https://dvcs.w3.org/hg/prov/raw-file/ae85f08dcda4/paq/prov-aq.html#locating-provenance-records

#g
--


On 07/02/2013 14:44, Tom De Nies wrote:
> Hi Paul, Graham,
>
> I'm confused about something in PROV-AQ.
>
> When linking through the HTTP Link header, you have the option to do this:
>
>> Link: http://example.org/example.provn; rel="
>> http://www.w3.org/ns/prov#hasProvenance"; anchor="
>> http://example.org/example.html"
>> Link: http://example.org/example2.provn; rel="
>> http://www.w3.org/ns/prov#hasProvenance"; anchor="
>> http://example.org/example.jpg"
>>
> Where one provenance resource describes the provenance of the html page,
> and another the provenance of a jpg included in the page.
> Another use case is the separate provenance of sections:
>
>> Link: http://example.org/example.provn; rel="
>> http://www.w3.org/ns/prov#hasProvenance"; anchor="
>> http://example.org/example.html#section1"
>> Link: http://example.org/example2.provn; rel="
>> http://www.w3.org/ns/prov#hasProvenance"; anchor="
>> http://example.org/example.html#section2"
>>
>
> However, when you use the <link> element in de <head> of the HTML, you can
> only do this:
>
>> <link rel="http://www.w3.org/ns/prov#hasProvenance" href="
>> http://example.org/example.provn">
>> <link rel="http://www.w3.org/ns/prov#hasProvenance" href="
>> http://example.org/example2.provn">
>> <link rel="http://www.w3.org/ns/prov#hasAnchor" href="
>> http://example.org/example.html">
>>
> if you would specify a second hadAnchor, you'd have no idea in which
> provenance resource the anchor is referred.
> Dito for RDF.
>
> Is the first HTTP example allowed or discouraged? And if it is allowed, how
> can it be done in HTML or RDF?
>
> This isn't intended as a blocking issue, just something I spotted when
> trying to link my personal page to its provenance.
>
> Regards,
> Tom
>

Received on Thursday, 7 February 2013 19:27:54 UTC